The LENS framework applied to 192 real-world settings shows moderate natural prompt distribution shifts cause 73% average performance loss in deployed LLMs, especially across user groups and regions.
Proceedings of the AAAI Conference on Artificial Intelligence , volume=
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CL 2years
2026 2representative citing papers
English-only safety alignment fails to transfer cross-lingually, while multilingual DPO training on the new RefusEU dataset improves safety across 12 European languages without degrading Global MMLU performance.
citing papers explorer
-
Measuring Distribution Shift in User Prompts and Its Effects on LLM Performance
The LENS framework applied to 192 real-world settings shows moderate natural prompt distribution shifts cause 73% average performance loss in deployed LLMs, especially across user groups and regions.
-
Multilingual Refusal Alignment for Safer Large Language Models
English-only safety alignment fails to transfer cross-lingually, while multilingual DPO training on the new RefusEU dataset improves safety across 12 European languages without degrading Global MMLU performance.