Shortcuts

BayesConv3d

class torch_uncertainty.layers.bayesian.BayesConv3d(in_channels, out_channels, kernel_size, stride=1, padding=0, dilation=1, prior_sigma_1=0.1, prior_sigma_2=0.002, prior_pi=1, mu_init=0.0, sigma_init=10.0, frozen=False, groups=1, bias=True, padding_mode='zeros', device=None, dtype=None)[source]

Bayesian Conv3d Layer with Gaussian mixture prior and Normal posterior.