What does logits mean in Tensorflow?
I'm checking out the loss functions of TensorFlow, and the majority of them have logits as a parameter. What does it mean?
I'm checking out the loss functions of TensorFlow, and the majority of them have logits as a parameter. What does it mean?
It is the value without probabilistic meaning, the inverse function of probabilistic mapping.
E.g. input argument of the sigmoid function.