sergeyprokudin

import keras.backend as K
import numpy as np


def gaussian_nll(ytrue, ypreds):
    """Keras implmementation of multivariate Gaussian negative loglikelihood loss function. 
    This implementation implies diagonal covariance matrix.
    
    Parameters
    ----------