What AI says about AQLM

AQLM is the official PyTorch implementation for extreme compression of large language models via additive quantization. It provides code and experiments to quantize, fine-tune, and convert LLMs using multiple codebooks to achieve smaller models with maintained accuracy. The project is actively maintained (e.g., release v1.1.7) and includes GPU-accelerated features such as arbitrary 8-dimensional codebooks to improve compressed model performance.