Request: Q8_0 mmproj
#1
by Zoont - opened
Hello, first of all thank you very much for uploading these quantized GGUFs as always, they are very helpful.
I have a friendly request which is can you consider adding Q8_0 mmproj files for multimodal models like this one? In my very limited testing they produce pretty much the same quality as F16 & BF16 versions, just like regular language model ggufs