Support new quantization formats
philpax opened this issue · comments
llama.cpp appears to have developed new quantization formats (K_S, K_M, K_L). We should investigate these and figure out what's required to support them.
Already opened another issue for this #301 😬
derp