Rate this Page

torch.Tensor.log_normal_#

Tensor.log_normal_(mean=1, std=2, *, generator=None)#

Fills self tensor with numbers samples from the log-normal distribution parameterized by the given mean μ\mu and standard deviation σ\sigma. Note that mean and std are the mean and standard deviation of the underlying normal distribution, and not of the returned distribution:

f(x)=1xσ2π e(lnxμ)22σ2f(x) = \dfrac{1}{x \sigma \sqrt{2\pi}}\ e^{-\frac{(\ln x - \mu)^2}{2\sigma^2}}