Comment by MyFirstSass

If this is "plug and play" can it be added to say llama.cpp and give ~3.67 speedup to existing models or is there some complication?