3 papers
cs.CL2026
Prism: Spectral-Aware Block-Sparse Attention
Xinghao Wang, Pengyu Wang, Xiaoran Liu +4
Block-sparse attention is promising for accelerating long-context LLM pre-filling, yet identifying relevant blocks efficiently remains a bottleneck. Existing methods typically empl…
cs.AI2026
AcuityBench: Evaluating Clinical Acuity Identification and Uncertainty Alignment
Robin Linzmayer, Georgianna Lin, Di Coneybeare +22
We introduce AcuityBench, a benchmark for evaluating whether language models identify the appropriate urgency of care from user medical presentations. Existing health benchmarks em…
cs.CY2026
LSSF: Safety Alignment for Large Language Models through Low-Rank Safety Subspace Fusion
Guanghao Zhou, Panjia Qiu, Cen Chen +4
The safety mechanisms of large language models (LLMs) exhibit notable fragility, as even fine-tuning on datasets without harmful content may still undermine their safety capabiliti…