🏢 Renmin University of China

Researchers developed SCAV, a novel framework to effectively reveal safety risks in LLMs by accurately interpreting their safety mechanisms. SCAV-guided attacks significantly improve attack success r…

Stable Diffusion’s text-to-image generation is sped up by 25% by removing text guidance after the initial shape generation, revealing that the [EOS] token is key to early-stage image construction.

Transformers’ in-context learning (ICL) is explained using representation learning, revealing its ICL process as gradient descent on a dual model and offering modifiable attention layers for enhanced …

S-MolSearch: a novel semi-supervised framework using 3D molecular data and contrastive learning achieves state-of-the-art in bioactive molecule search, outperforming existing methods.

Researchers unveil text-level graph injection attacks, revealing a new vulnerability in GNNs and highlighting the importance of text interpretability in attack success.

SVD-based weight pruning surprisingly boosts in-context learning in large language models, especially when applied to deeper layers, offering a novel approach to model compression and efficiency.

A novel conjugate Bayesian two-step change point detection method for Hawkes processes, CoBay-CPD, achieves higher accuracy and efficiency by employing data augmentation for improved dynamic event mod…

CausalStock: A novel framework for accurate news-driven multi-stock movement prediction, using lag-dependent causal discovery and LLMs for enhanced noise reduction and explainability.

Householder Reflection Adaptation (HRA) bridges low-rank and orthogonal LLM adaptation, achieving superior performance with fewer parameters than existing methods. By using a chain of Householder refl…