Moe at Scale: Making Sparse Models Fast on Real Hardware

(cerebras.ai)

2 points | by frozenseven 6 hours ago ago