With Python and NumPy getting lots of exposure lately ... 0.26 = 0.7229 The softmax function is perhaps best explained by example: divisor = exp(0.6911) + exp(0.7229) = 1.9960 + 2.0604 = 4.0563 output ...