Huawei’s Computing Systems Lab in Zurich has introduced a new open-source quantization method for large language models (LLMs) aimed at reducing memory demands without sacrificing output quality.