5 papers
CustomX: Unified Character, Action, and Scene Customization in Video World Models
Yitong Wang, Fangyun Wei, Hongyang Zhang +2
Recent advances in world models have greatly enhanced interactive environment simulation. Existing methods mainly fall into two categories: (1) static world generation models, whic…
Spatia: Video Generation with Updatable Spatial Memory
Jinjing Zhao, Fangyun Wei, Zhening Liu +3
Existing video generation models struggle to maintain long-term spatial and temporal consistency due to the dense, high-dimensional nature of video signals. To overcome this limita…
From Virtual Games to Real-World Play
Wenqiang Sun, Fangyun Wei, Jinjing Zhao +5
We introduce RealPlay, a neural network-based real-world game engine that enables interactive video generation from user control signals. Unlike prior works focused on game-style v…
EAGLE-3: Scaling up Inference Acceleration of Large Language Models via Training-Time Test
Yuhui Li, Fangyun Wei, Chao Zhang +1
The sequential nature of modern LLMs makes them expensive and slow, and speculative sampling has proven to be an effective solution to this problem. Methods like EAGLE perform auto…
EAGLE: Speculative Sampling Requires Rethinking Feature Uncertainty
Yuhui Li, Fangyun Wei, Chao Zhang +1
Autoregressive decoding makes the inference of Large Language Models (LLMs) time-consuming. In this paper, we reconsider speculative sampling and derive two key observations. First…