pub fn load_ssm(
store: &WeightStore,
layer_prefix: &str,
gpu: &dyn GpuBackend,
variant: Nvfp4Variant,
qctx: QuantizeCtx,
config: &ModelConfig,
) -> Result<SsmWeights>Expand description
2026-09-25: Load the {layer_prefix}.linear_attn weights of a linear-attention layer.