pub fn elemwise_erf_backward<T: Triton, D: Float, const BLOCK_SIZE: i32>(
dy_ptr: T::Pointer<D>,
x_ptr: T::Pointer<D>,
dx_ptr: T::Pointer<D>,
n_elements: i32,
)where
T::I32Tensor: Tensor<i32, 1> + Comparison<i32, BoolTensor = T::BoolTensor>,
T::Pointer<D>: AddOffsets<i32, 1, T::I32Tensor, Output = T::Tensor<T::Pointer<D>>>,Expand description
Backward: dx = 2/sqrt(pi) * exp(-x^2) * dy where 2/sqrt(pi) ~= 1.1283791670955126