3 ms·
why 27b vs 35b? Is MoE that much worse for coding?
by htrp 4mo ago
why 27b vs 35b? Is MoE that much worse for coding?
- electronsoup 4mo agoYeah MoE is a little worse for the same size, but you can often run bigger MoEs at respectable speeds even on cpu ram offload. The dense models really need to be 100% vram
- amarshall 4mo agoCan take the geometric mean of total and active parameters of MoE to get approximate equivalent quality to dense model params. So sqrt(35*10)≈18.7. The trade-off of MoE is that it is worse but faster for the same total size.