You can run a 120B large model without 60GB of memory! An overseas user successfully linked a phone, Mac, and Windows to achieve it.
Generally speaking, if you want to run a 120-billion-parameter large language model locally, the standard approach is to first get ...