4 papers
Beyond Layers: Position-Resolved Gradient Conflict and Position-Aware Modulation for Unified Multimodal Models
Shuyang Jiang, Fucheng Deng, Yuchuan Luo +1
Unified multimodal models (UMMs) train image understanding and autoregressive image generation on shared parameters, and the two objectives are known to interfere. Existing diagnos…
SeMob: Semantic Synthesis for Dynamic Urban Mobility Prediction
Runfei Chen, Shuyang Jiang, Wei Huang
Human mobility prediction is vital for urban services, but often fails to account for abrupt changes from external events. Existing spatiotemporal models struggle to leverage textu…
Linear Attention via Orthogonal Memory
Jun Zhang, Shuyang Jiang, Jiangtao Feng +2
Efficient attentions have greatly improved the computational efficiency of Transformers. However, most existing linear attention mechanisms suffer from an \emph{efficiency degradat…
Attentive Multi-Layer Perceptron for Non-autoregressive Generation
Shuyang Jiang, Jun Zhang, Jiangtao Feng +2
Autoregressive~(AR) generation almost dominates sequence generation for its efficacy. Recently, non-autoregressive~(NAR) generation gains increasing popularity for its efficiency a…