mlabonne/Meta-Llama-3.1-8B-Instruct-abliterated-GGUF · model optimization to decrease sizes of quants

Hi, i try this model it works fine! Thanks again, i think about making backup of llama 3.1 8b model on DvD disk 4.7GB that cheap,common and long lasting backup storage i have disks that burn in 2008 ,md5 of files are ok, the problem is that Q4_K_S is 4.72GB, Q4_K_M quant is 4.92 GB and of course not fit on dvd, lower quants 4 _0 is 4.69gb almost fit in dvd, but q4_0 and below made model more dumb and loss of quality , there any methods to optimize model before turn in quants to decrease size of model and quants ??