Geometry-Lite decomposes LLM safety detection into layer-wise margin geometries and finds that persistent boundary positions, not layer-to-layer drift, drive most detection performance across nine models and seven benchmarks.
Pku-saferlhf: Towards multi-level safety alignment for llms with human preference
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Geometry-Lite: Interpretable Safety Probing via Layer-Wise Margin Geometry
Geometry-Lite decomposes LLM safety detection into layer-wise margin geometries and finds that persistent boundary positions, not layer-to-layer drift, drive most detection performance across nine models and seven benchmarks.