The Extreme LLM Compression Evolution: From QuIP to AQLM With PV-Tuning | HackerNoonLarge language models can be compressed from 16 to 2 bits using methods like AQLM and PV-Tuning, enabling significant model size reduction.