7 papers
CuMA: Aligning LLMs with Sparse Cultural Values via Demographic-Aware Mixture of Adapters
Ao Sun, Xiaoyu Wang, Zhe Tan +4
As Large Language Models (LLMs) serve a global audience, alignment must transition from enforcing universal consensus to respecting cultural pluralism. We demonstrate that dense mo…
The Need for a Socially-Grounded Persona Framework for User Simulation
Pranav Narayanan Venkit, Yu Li, Yada Pruksachatkun +1
Synthetic personas are widely used to condition large language models (LLMs) for social simulation, yet most personas are still constructed from coarse sociodemographic attributes…
SilentStriker:Toward Stealthy Bit-Flip Attacks on Large Language Models
Haotian Xu, Qingsong Peng, Jie Shi +3
The rapid adoption of large language models (LLMs) in critical domains has spurred extensive research into their security issues. While input manipulation attacks (e.g., prompt inj…
Linear Preference Optimization: Decoupled Gradient Control via Absolute Regularization
Rui Wang, Qianguo Sun, Chao Song +4
DPO (Direct Preference Optimization) has become a widely used offline preference optimization algorithm due to its simplicity and training stability. However, DPO is prone to overf…
GPAS: Accelerating Convergence of LLM Pretraining via Gradient-Preserving Activation Scaling
Tianhao Chen, Xin Xu, Zijing Liu +12
Modern Large Language Models, such as the LLaMA, Qwen and DeepSeek series, predominantly adopt the Pre-LayerNorm (Pre-LN) Transformer architecture. While being stable during pretra…
InverTune: Removing Backdoors from Multimodal Contrastive Learning Models via Trigger Inversion and Activation Tuning
Mengyuan Sun, Yu Li, Yuchen Liu +2
Multimodal contrastive learning models like CLIP have demonstrated remarkable vision-language alignment capabilities, yet their vulnerability to backdoor attacks poses critical sec…