Tongxin Bai — accepted AI-conference papers

2026

SparVAR: Exploring Sparsity in Visual AutoRegressive Modeling for Training-Free Acceleration

Zekun Li, Ning Wang, Tongxin Bai, Changwang Mei, Peisong Wang, Shuang Qiu +1

CVPR 2026

Visual AutoRegressive (VAR) modeling has garnered significant attention for its innovative next-scale prediction paradigm. However, mainstream VAR paradigms attend to all tokens across historical scales at each autoregressive step. As the next scale resolution grows, the computational complexity of

Cited by 0 Source ↗code ↗Scholar ↗

2025

MVA: Linear Attention with High-order Query-Keys Integration and Multi-level Vocabulary Decomposition

wang ning, Zekun Li, Tongxin Bai, Man Yao, Zhen Qin, Guoqi Li

ICML 2025poster

Linear attention offers the advantages of linear inference time and fixed memory usage compared to Softmax attention. However, training large-scale language models with linear attention from scratch remains prohibitively expensive and exhibits significant performance gaps compared to Softmax-based…

Cited by 0 Source ↗PDF ↗Scholar ↗