Lixuan He — accepted AI-conference papers

2026

A Flat Vocabulary or a Rich Hierarchy? Re-introducing Intrinsic Structure Transforms the Autoregressive Image Generation

ICML 2026poster

Autoregressive (AR) models have shown great promise in image generation, yet they face a fundamental inefficiency stemming from their core component: a vast, unstructured vocabulary of visual tokens. By treating tokens as a flat set, standard models overlook the manifold structure where geometric pr…

Cited by 0 Source ↗Scholar ↗

2026

Let Features Decide Their Own Solvers: Hybrid Feature Caching for Diffusion Transformers

Shikang Zheng, Guantao Chen, Qinming Zhou, Yuqi Lin, Lixuan He, Chang Zou +3

ICLR 2026oral

Diffusion Transformers (DiTs) offer state-of-the-art fidelity in image and video synthesis, but their iterative sampling process remains a major bottleneck due to the high cost of transformer forward passes at each timestep. To mitigate this, feature caching has emerged as a training-free accelerati…

Cited by 0 Source ↗Scholar ↗