9 papers
MergeTok: Unified Continuous and Discrete Visual Tokenization via Token Merging
Luyuan Zhang, Siyuan Li, Zedong Wang +7
Most visual tokenizers for image generation are bifurcated into two families with complementary limitations: continuous VAEs offer high-fidelity reconstruction but suffer from dens…
RankE: End-to-End Post-Training for Discrete Text-to-Image Generation with Decoder Co-Evolution
Siyong Jian, Siyuan Li, Luyuan Zhang +5
Discrete autoregressive (AR) text-to-image (T2I) models pair a VQ tokenizer with an AR policy, and current post-training pipelines optimize only the policy while keeping the VQ dec…
CARE-Edit: Condition-Aware Routing of Experts for Contextual Image Editing
Yucheng Wang, Zedong Wang, Yuetong Wu +2
Unified diffusion editors often rely on a fixed, shared backbone for diverse tasks, suffering from task interference and poor adaptation to heterogeneous demands (e.g., local vs gl…
Rep-MTL: Unleashing the Power of Representation-level Task Saliency for Multi-Task Learning
Zedong Wang, Siyuan Li, Dan Xu
Despite the promise of Multi-Task Learning in leveraging complementary knowledge across tasks, existing multi-task optimization (MTO) techniques remain fixated on resolving conflic…
MogaNet: Multi-order Gated Aggregation Network
Siyuan Li, Zedong Wang, Zicheng Liu +6
By contextualizing the kernel as global as possible, Modern ConvNets have shown great potential in computer vision tasks. However, recent progress on multi-order game-theoretic int…
Taming LLMs by Scaling Learning Rates with Gradient Grouping
Siyuan Li, Juanxi Tian, Zedong Wang +4
Training large language models (LLMs) poses challenges due to their massive scale and heterogeneous architectures. While adaptive optimizers like AdamW help address gradient variat…