Tag
This paper proposes D-SCAN, a detection framework for RAG poisoning that monitors attention collapse dynamics in language model generations, showing effectiveness against adversarial attacks even when outputs appear benign.
This paper identifies a powerful space-based GNSS interference source over Europe, Greenland, and Canada as a constellation of Russian early warning satellites in Molniya orbits, based on data from 2019 to 2026.
LaRA is a layer-wise representation analysis framework that detects data contamination in RL post-trained LLMs by measuring geometric deviations across model layers, outperforming output-level baselines.