I packed 16 GB of GGUF quants into 1.8 GB, losslessly

v3ss0n

for disk only

sarjann

Seems pretty useful, often hopping between different quants. I wonder if this would work for different "branches" of a model. E.g. qwen 3.6 35b regular vs abliterated.

VASTL

No offense, but do you really you "cracked the code" if multi-billion/trillion companies cant achieve that? I heavily doubt, you have losless quality with those 25 commits on GH, most likely vibecoded.