Skip to main content

elemwise_clip_backward

teeny_kernels::nn::tensor::elemwise_binary

Function elemwise_clip_backward 

Source 
pub fn elemwise_clip_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,
    min_val: f32,
    max_val: f32,
)
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: pass dy through only where x was in [min_val, max_val]