pub fn k5_ssd_chunk_scan(
da_cumsum_bhnl: Tensor<4>,
dt_discretized_bhnl: Tensor<4>,
x_bnlhp: Tensor<5>,
c_bnlhr: Tensor<5>,
cb_bnhll: Tensor<5>,
chunk_input_state_bnhpr: Tensor<5>,
d_h: Tensor<1>,
) -> Tensor<5>Expand description
Based on the Kernel 5 Triton reference _chunk_scan_fwd_kernel (ssd_chunk_scan.py).
Returns:
- y_bnlhp
[final output]