Learning to count in Binary

Objective

Make a Neural Network learn how to count. In the example below, the network receives a binary number of length 3 and computes the next number.

So for 010 the output must be 011. For 011 the output is 100.

Live evolution of the network

I0, I1, I2 - inputs
B6 - bias
O3, O4, O5 - outputs

Result

(Left - output | Right - input)