As title says. For example if my tensor represents a probability, how do I, in a simple way, ensure it’s always in the range of 0-1 through backpropagations?(not by implementing situation-specific tricks like logit transfer)
As title says. For example if my tensor represents a probability, how do I, in a simple way, ensure it’s always in the range of 0-1 through backpropagations?(not by implementing situation-specific tricks like logit transfer)