2 ms·The paper is about vector quantization, which affects KV cache not model weights/sizes.by rohansood15 7mo agoThe paper is about vector quantization, which affects KV cache not model weights/sizes.