decay_rate

  • null (default)


The learning_rate is a parameter that determines how much the update step affects the current value of the weights while the decay_rate is an additional term in a weight update rule that exponentially reduces a weight to zero.