uber/ludwig
Model always predicts 1 or 0 on multiclass classification problemI have verified that I have the same issue while using a similar dataset on a related binary classification problem. 99% of the output probabilities are 1 or 0. I will work on producing a reproducible example. |