dp
3 items · generated by scribe fromstdlib/dp.fd· synced 2026-11-15
stdlib/dp3 items
Contents
- dp_noise (function)
- gaussian_noise (function)
- dropout (function)
dp_noise
Add Gaussian noise to a tensor for differential privacy. sigma = sqrt(2 * epsilon * sensitivity^2 / num_samples)
pub fn dp_noise(t: Tensor, sigma: f64) -> Tensor {Parameters:
t:Tensorsigma:f64
Returns: Tensor
Defined at line 10
gaussian_noise
Generate Gaussian noise tensor with given shape and sigma.
pub fn gaussian_noise(shape: Vec<i64>, sigma: f64) -> Tensor {Parameters:
shape:Vec<i64>sigma:f64
Returns: Tensor
Defined at line 15
dropout
Dropout: randomly zero elements with probability p, scale survivors.
pub fn dropout(t: Tensor, p: f64) -> Tensor {Parameters:
t:Tensorp:f64
Returns: Tensor
Defined at line 20
Generated by scribe, the Fides documentation generator