Xin Li

Interactive image refocusing is now 462x faster, achieving sub-second edits by separating generative restoration from efficient blur rendering.

Autonomous ML engineering agents can achieve frontier-model performance by learning from long experimental trajectories using outcome-guided supervision and windowed reinforcement learning.

LATCH accelerates diffusion language model decoding by up to 17.8x without compromising accuracy, redefining how we approach early exits in generative tasks.