Comment Comment by preya2k Are you on Strix Halo? If so, which llama.cpp backend are you using? ROCm on Strix Halo still seems incompatible with MTP models. parent preya2k 3 months ago
