Can't wait to test! This should significantly boost TPS! Now we just need more llama cpp optimizations to be merged in!
