silu(gate(x))·up(x) for exactly one activation row, gate and up at the same
width: the per-row kernel. codebook applies at the measured shape; layers
pass the choice they resolved at construction (qualifiedRepresentations).
Borrows x and the weights; returns owned lazy output with x's leading dimensions.
silu(gate(x))·up(x) for exactly one activation row, gate and up at the same width: the per-row kernel.
codebookapplies at the measured shape; layers pass the choice they resolved at construction (qualifiedRepresentations). Borrows x and the weights; returns owned lazy output with x's leading dimensions.