pub fn reptile_update_simd(target: &mut [f32], source: &[f32], lr: f32)Expand description
SIMD-friendly Reptile update: target[i] += lr * (source[i] - target[i])
Processes 8 f32s at a time via chunks_exact, giving LLVM a clear
vectorization opportunity (256-bit AVX2 = 8 x f32).
ยงPanics
Panics if target.len() != source.len().