#activation patching
Activation Patching: 2 AI articles covering activation patching news, analysis, and research
Articles
A Mechanistic Study of AI-Text Detection Neurons in Frozen BERT:⭐9
This study identifies sparse AI-text detection neurons in frozen BERT via sparse probing and activation patching on RAID, revealing a layer-12 alignment footpri...
Subliminal Prompting Beyond Static Geometry: Causal Depth and⭐9
Subliminal prompting study finds fixed output-vector similarity weakens at scale while causal control and a number-width length confound reshape how hidden trai...