Sheng Zhang

Autoregressive video models can now perform explicit, goal-directed reasoning by jointly predicting the final outcome and intermediate steps, achieving state-of-the-art performance with 75% fewer training steps.

Exact closed-form solutions for non-diagonal no-jump operators eliminate the need to simulate fault-free paths in amplitude-damped QRAM, accelerating classical noise modeling by up to 285×.

GigaPath-Flash and GigaTIME-Flash deliver state-of-the-art pathology analysis with drastically reduced computational costs, enabling broader clinical adoption.