Module flowcon.nn.nets.spectral_norm

Functions

def asym_squash(x)

def leaky_elu(x, a=0.3)

def projmax_(v)

Inplace argmax on absolute value.

def scaled_spectral_norm(module: torch.nn.modules.module.Module, domain, codomain, coeff=0.97, name: str = 'weight', n_power_iterations: int = 1, eps: float = 1e-12, dim: Optional[int] = None) ‑> torch.nn.modules.module.Module

Applies spectral normalization to a parameter in the given module. Calls custom normalization modules instead of default pytorch ones.

def vector_norm(x, p)