pub fn yolo_bce_cls_loss_forward<T: Triton, D: Float, const BLOCK_N: i32>(
pred_ptr: T::Pointer<D>,
target_ptr: T::Pointer<D>,
loss_ptr: T::Pointer<D>,
N: i32,
C: i32,
)Expand description
BCE classification loss forward: class logits + soft targets → per-anchor loss.
Grid: cdiv(N, BLOCK_N) — one CTA per anchor tile.