scarlet2.nn

Neural network priors

Classes

ScorePrior(model, shape, *args, **kwargs)

Score-matching neural network to represent the prior distribution

Functions

calc_grad(x, model)	Calculates the gradient of the log-prior using the ScoreNet model chosen
pad_back(x, pad)	Removes the zero-padding from the input image
pad_fwd(x, model_shape)	Zero-pads the input image to the model size
vgrad(f, x)