10 papers
Smooth Scaling Laws Hide Stepwise Token Learning
Pingjie Wang, Zechen Hu, Peiru Yang +2
Language model loss follows remarkably regular scaling laws over model and data size, yet it remains unclear why the aggregate loss should exhibit a power-law form. Existing explan…
Knowledge Poisoning Attacks on Medical Multi-Modal Retrieval-Augmented Generation
Peiru Yang, Haoran Zheng, Tong Ju +6
Retrieval-augmented generation (RAG) is a widely adopted paradigm for enhancing LLMs in medical applications by incorporating expert multimodal knowledge during generation. However…
Igusa-Todorov properties of recollements of abelian categories
Peiru Yang, Yajun Ma, Yu-Zhe Liu
In this paper, we investigate the behavior of Igusa-Todorov properties under recollements of abelian categories. In particular, we study how the Igusa-Todorov distances of the cate…
LiveSecBench: A Dynamic and Event-Driven Safety Benchmark for Chinese Language Model Applications
Yudong Li, Peiru Yang, Feng Huang +18
We introduce LiveSecBench, a continuously updated safety benchmark specifically for Chinese-language LLM application scenarios. LiveSecBench constructs a high-quality and unique da…
Black-Box Membership Inference Attack for LVLMs via Prior Knowledge-Calibrated Memory Probing
Jinhua Yin, Peiru Yang, Chen Yang +5
Large vision-language models (LVLMs) derive their capabilities from extensive training on vast corpora of visual and textual data. Empowered by large-scale parameters, these models…
RedNote-Vibe: A Dataset for Capturing Temporal Dynamics of AI-Generated Text in Lifestyle Social Media
Yudong Li, Yufei Sun, Peiru Yang +7
We introduce RedNote-Vibe, a dataset spanning five years (pre-LLM to July 2025) sourced from lifestyle platform RedNote (Xiaohongshu), capturing the temporal dynamics of content cr…
Enhancing Watermarking Quality for LLMs via Contextual Generation States Awareness
Peiru Yang, Xintian Li, Wanchun Ni +6
Recent advancements in watermarking techniques have enabled the embedding of secret messages into AI-generated text (AIGT), serving as an important mechanism for AIGT detection. Ex…
MrM: Black-Box Membership Inference Attacks against Multimodal RAG Systems
Peiru Yang, Jinhua Yin, Haoran Zheng +7
Multimodal retrieval-augmented generation (RAG) systems enhance large vision-language models by integrating cross-modal knowledge, enabling their increasing adoption across real-wo…
HeteRAG: A Heterogeneous Retrieval-augmented Generation Framework with Decoupled Knowledge Representations
Peiru Yang, Xintian Li, Zhiyang Hu +8
Retrieval-augmented generation (RAG) methods can enhance the performance of LLMs by incorporating retrieved knowledge chunks into the generation process. In general, the retrieval…
Evidencing Unauthorized Training Data from AI Generated Content using Information Isotopes
Qi Tao, Yin Jinhua, Cai Dongqi +10
In light of scaling laws, many AI institutions are intensifying efforts to construct advanced AIs on extensive collections of high-quality human data. However, in a rush to stay co…