Function mla_decode_lse_f32 

pub fn mla_decode_lse_f32(
    query: &[f32],
    query_pe: &[f32],
    key_value: &[f32],
    key_pe: &[f32],
    batch: usize,
    key_len: usize,
    heads: usize,
    head_dim: usize,
    pe_dim: usize,
    scale: f32,
) -> (Vec<f32>, Vec<f32>)