strange, why is Q3K_XL even smaller than Q3K_M?

#10
by X5R - opened

is it not only Q8_0 on some tensors comparing to Q3K_M?

the quant provided by others is larger than q3k_m

Unsloth AI org

is it not only Q8_0 on some tensors comparing to Q3K_M?

Its normal because some layers don't have to be higher precision

Sign up or log in to comment