flashinfer.cake_causal_conv1d_nondet_train.cake_causal_conv1d_nondet_train_backward_workspace_bytes

flashinfer.cake_causal_conv1d_nondet_train.cake_causal_conv1d_nondet_train_backward_workspace_bytes(S: int, C: int, device: device, *, bwd_grid_mult: int = 1) → int

Return the backward workspace size in bytes.

Parameters:
  • S (int) – Packed token count.

  • C (int) – Channel count.

  • device (torch.device) – CUDA device that will execute the kernel.

  • bwd_grid_mult (int, default=1) – Backward grid multiplier to validate for the selected backend.

Returns:

Bytes of the caller-owned backward workspace. The size depends on C and the target device, not on S.

Return type:

int