Fix 4-bit quantization for weight matrices not divisible by blocksize#1884

matthewdouglas merged 3 commits intobitsandbytes-foundation:mainbitsandbytes-foundation/bitsandbytes:mainfrom

Abdennacer-Badaoui:triton-enhancementsAbdennacer-Badaoui/bitsandbytes:triton-enhancementsCopy head branch name to clipboard