7 papers
ERA: Entropy-Guided Visual Token Pruning with Rectified Attention for Efficient MLLMs
Yuhao Wang, Mu Qiao, Haiwen Diao +5
Multimodal Large Language Models (MLLMs) incur prohibitive inference costs due to long visual token sequences. Training-free visual token reduction provides an efficient solution.…
From Pixels to Words -- Towards Native One-Vision Models at Scale
Haiwen Diao, Jiahao Wang, Penghao Wu +18
Current vision-language models (VLMs) typically stitch together separate image encoders and language decoders via multi-stage alignment, a modular framework that inevitably fragmen…
VISTA-Bench: Do Vision-Language Models Really Understand Visualized Text as Well as Pure Text?
Qing'an Liu, Juntong Feng, Yuhao Wang +6
Vision-Language Models (VLMs) have achieved impressive performance in cross-modal understanding across textual and visual inputs, yet existing benchmarks predominantly focus on pur…
EVEv2: Improved Baselines for Encoder-Free Vision-Language Models
Haiwen Diao, Xiaotong Li, Yufeng Cui +6
Existing encoder-free vision-language models (VLMs) are rapidly narrowing the performance gap with their encoder-based counterparts, highlighting the promising potential for unifie…
End-to-End Vision Tokenizer Tuning
Wenxuan Wang, Fan Zhang, Yufeng Cui +5
Existing vision tokenization isolates the optimization of vision tokenizers from downstream training, implicitly assuming the visual tokens can generalize well across various tasks…
Autoregressive Video Generation without Vector Quantization
Haoge Deng, Ting Pan, Haiwen Diao +6
This paper presents a novel approach that enables autoregressive video generation with high efficiency. We propose to reformulate the video generation problem as a non-quantized au…