7 papers
Sockpuppetting: Jailbreaking LLMs by Combining Prefilling with Optimization
Asen Dotsinski, Panagiotis Eustratiadis
Prefill attacks are an effective and low-cost jailbreaking method, as they directly insert an acceptance sequence (e.g., "Sure, here is...") at the start of an LLM's output and lea…
On the Robustness of LLM-Based Dense Retrievers: A Systematic Analysis of Generalizability and Stability
Yongkang Li, Panagiotis Eustratiadis, Yixing Fan +1
Decoder-only large language models (LLMs) are increasingly replacing BERT-style architectures as the backbone for dense retrieval, achieving substantial performance gains and broad…
Spectral Tempering for Embedding Compression in Dense Passage Retrieval
Yongkang Li, Panagiotis Eustratiadis, Evangelos Kanoulas
Dimensionality reduction is critical for deploying dense retrieval systems at scale, yet mainstream post-hoc methods face a fundamental trade-off: principal component analysis (PCA…
Unsupervised Corpus Poisoning Attacks in Continuous Space for Dense Retrieval
Yongkang Li, Panagiotis Eustratiadis, Simon Lupart +1
This paper concerns corpus poisoning attacks in dense information retrieval, where an adversary attempts to compromise the ranking performance of a search algorithm by injecting a…
Information Leakage of Sentence Embeddings via Generative Embedding Inversion Attacks
Antonios Tragoudaras, Theofanis Aslanidis, Emmanouil Georgios Lionis +2
Text data are often encoded as dense vectors, known as embeddings, which capture semantic, syntactic, contextual, and domain-specific information. These embeddings, widely adopted…
Reproducing HotFlip for Corpus Poisoning Attacks in Dense Retrieval
Yongkang Li, Panagiotis Eustratiadis, Evangelos Kanoulas
HotFlip is a topical gradient-based word substitution method for attacking language models. Recently, this method has been further applied to attack retrieval systems by generating…