pub fn ffn_gateup_fused_parts(hidden: usize, inter: usize) -> (usize, usize)Expand description
ffn_gateup_fused_bytes split into (weight bytes, scale-grid bytes) —
the loader adopts the two buffers separately, so it needs the terms rather
than the sum, and taking them from here is what keeps the prediction and
the tally one arithmetic.