Congratulations to Kangbo, Yiyang and Wentao. We will present three accelerator designs for Multimodal LLM, 3D Gaussian Splatting, and LLM inference with 3D stacked memory.