master-8ab8ba6
8ab8ba62 · llama : prevent usage of k-quants when tensor size is not a multiple of 256 (#1921) ·