pub fn hc_post_site(
gpu: &dyn GpuBackend,
kernel: KernelHandle,
hc: &HcWeights,
block_out: DevicePtr,
residual: DevicePtr,
post: DevicePtr,
comb: DevicePtr,
out: DevicePtr,
num_tokens: u32,
hidden_size: u32,
stream: u64,
) -> Result<()>Expand description
Inject the block output back into every stream. out may alias
residual.
Takes the whole HcWeights rather than a site, because NEITHER variant’s
hc_post reads site weights — Sinkhorn consumes the comb its hc_pre
emitted, low-rank the injection vector — so the layer’s variant is the
only thing being selected on.