29 citations · 49 across the 14 of their papers we have counts for
7 papers · 1 filter
HalluciDoctor: Mitigating Hallucinatory Toxicity in Visual Instruction Data
Qifan Yu, Juncheng Li, Longhui Wei +6
Multi-modal Large Language Models (MLLMs) tuned on machine-generated instruction-following data have demonstrated remarkable performance in various multi-modal understanding and ge…
Degeneration-Tuning: Using Scrambled Grid shield Unwanted Concepts from Stable Diffusion
Zixuan Ni, Longhui Wei, Jiacheng Li +3
Owing to the unrestricted nature of the content in the training data, large text-to-image diffusion models, such as Stable Diffusion (SD), are capable of generating images with pot…
SDDM: Score-Decomposed Diffusion Models on Manifolds for Unpaired Image-to-Image Translation
Shikun Sun, Longhui Wei, Junliang Xing +2
Recent score-based diffusion models (SBDMs) show promising results in unpaired image-to-image translation (I2I). However, existing methods, either energy-based or statistically-bas…
Continual Vision-Language Representation Learning with Off-Diagonal Information
Zixuan Ni, Longhui Wei, Siliang Tang +2
Large-scale multi-modal contrastive learning frameworks like CLIP typically require a large amount of image-text samples for training. However, these samples are always collected c…
Learning Transferable Pedestrian Representation from Multimodal Information Supervision
Liping Bao, Longhui Wei, Xiaoyu Qiu +3
Recent researches on unsupervised person re-identification~(reID) have demonstrated that pre-training on unlabeled person images achieves superior performance on downstream reID ta…
Lformer: Text-to-Image Generation with L-shape Block Parallel Decoding
Jiacheng Li, Longhui Wei, ZongYuan Zhan +4
Generative transformers have shown their superiority in synthesizing high-fidelity and high-resolution images, such as good diversity and training stability. However, they suffer f…
Gradient-Regulated Meta-Prompt Learning for Generalizable Vision-Language Models
Juncheng Li, Minghe Gao, Longhui Wei +7
Prompt tuning, a recently emerging paradigm, enables the powerful vision-language pre-training models to adapt to downstream tasks in a parameter -- and data -- efficient way, by l…