I get a model result with all Wi = NaN

[makailove123]

I trained a model with 2500000 samples and 900 features. But the result Wi were all NaN. I randomly chose 1000 samples from the dataset, and the reault seems ok.
Does it have limitation on the size of dataset or feature values?

[takun2s]

No, there is no limitation of the data size .
You could try to normalize the data .

[qf6101]

I encountered a similar overflow problem with SoftMax. And I fix the code as follows.

val margins = ws.map { w =>
  VectorUtil.wxpb(w,x,1d)
}
val maxMargin = margins.max
val adjustedMargins = margins.map { margin =>
  margin - maxMargin
}
val mes = adjustedMargins.map {...