gumbel_like

gumbel_like(
    x: &Tensor<T>,
    mu: T,
    beta: T
) -> Result<Tensor<T>, TensorError>

Same as gumbel but the shape will be based on x. Creates a Tensor with values drawn from a Gumbel distribution with location parameter mu and scale parameter beta.

Parameters:

x: Input Tensor to derive the shape from

mu: Location parameter (μ) of the Gumbel distribution.

beta: Scale parameter (β) of the Gumbel distribution. Must be positive.

Returns:

Tensor with type T containing random values from the Gumbel distribution.

Examples:

use hpt::{error::TensorError, ops::Random, Tensor};

fn main() -> Result<(), TensorError> {
    // Create an initial tensor
    let x = Tensor::<f32>::randn(&[10, 10])?;
    
    // Create a new tensor with same shape as x but with Gumbel distribution
    let g = x.gumbel_like(0.0, 1.0)?;
    println!("{}", g);
    Ok(())
}

Backend Support