Maybe something wrong in updating lambda

[DingJianhao]

In https://github.com/PotatoThanh/ADMM-NeuralNetworks/blob/master/python/ADMM_NN.py#L197, your code goes like 'lambda = lambda_update'.
But as implied in paper, this code should goes like 'lambda += lambda'.
After change that, the losses are going up.
Have I misunderstood something?