Skip to contents

Element-wise exp(x) - 1, more accurate for small x.

Usage

nvl_expm1(operand)

Arguments

operand

(tensorish)
Tensorish value of data type floating-point.

Value

tensorish
Has the same shape and data type as the input. It is ambiguous if the input is ambiguous.

Implemented Rules

  • stablehlo

  • backward

See also

Examples

jit_eval({
  x <- nv_tensor(c(0, 0.001, 1))
  nvl_expm1(x)
})
#> AnvilTensor
#>  0.0000
#>  0.0010
#>  1.7183
#> [ CPUf32{3} ]