Hui-Ling Zhen

Sparse by Command: Task-Conditional Compute Skipping for Multi-Task Inference Accelerators

Afzal Ahmad, Gaoyu Mao, Shoubo Hu, Hui-Ling Zhen, Mingxuan Yuan, Xinyu Chen, et al.

Multi-task inference models share a single backbone across diverse tasks, yet execute identical computation regardless of which task is active - wasting energy and cycles on task-irrelevant operations. We observe that the task command, typically available before inference begins,…Read more