A continuation of that trend (that, for instance, a 65B model would quantize better than a 33B one) was popular belief, but the k-quants dev recently found that

A continuation of that trend (that, for instance, a 65B model would quantize better than a 33B one) was popular belief, but the k-quants dev recently found that LLaMA 7B-65B all quantize very similarly: https://github.com/ggerganov/llama.cpp/pull/1684 https://github.com/ggerganov/llama.cpp/pull/1684