Function linear_packed 

pub fn linear_packed(
    x: &[f32],
    weight: &[f32],
    bias: Option<&[f32]>,
    m: usize,
    k: usize,
    n: usize,
    out: &mut [f32],
)

out[m, n] = x[m, k] @ weight[n, k]^T + bias[n].

weight is the checkpoint’s native [out_channels, in_channels] layout — each output row contiguous — so no transpose is ever materialized, matching the project’s one GEMM contract.

Panics

If the slice lengths disagree with m, k, n.