1-Bit AI and the Collapse of the Scale Paradigm

1/ A striking claim out of @Caltech: a 1-bit large language model that achieves massive compression without performance loss. (https://www.wsj.com/cio-journal/caltech-researchers-claim-radical-compres…