Yuting Wei

Sampling complexity can be reduced to logarithmic factors, challenging the conventional belief that it scales linearly with ambient dimension.

RLVR's success in long-horizon reasoning hinges on a smooth difficulty spectrum, where mastering easier sub-problems unlocks the ability to tackle harder ones, avoiding frustrating grokking plateaus.

Discrete diffusion models can now sample high-dimensional data with provable sublinear convergence rates, adapting to hidden structure without prior knowledge.