MulTCIM: Digital Computing-in-Memory-Based Multimodal Transformer Accelerator With Attention-Token-Bit Hybrid Sparsity

Fengbin Tu, Zihan Wu 0006, Yiqi Wang 0005, Weiwei Wu, Leibo Liu, Yang Hu 0001, Shaojun Wei, Shouyi Yin. MulTCIM: Digital Computing-in-Memory-Based Multimodal Transformer Accelerator With Attention-Token-Bit Hybrid Sparsity. J. Solid-State Circuits, 59(1):90-101, January 2024. [doi]

Abstract

Abstract is missing.