Reliable concept presence in transformers is concentrated in the extreme high-activation tail of in-concept tokens; thresholding that tail improves concept detection and localization.
Unmasking and quantifying racial bias of large language models in medical report generation.Communications Medicine, 4(1), September 2024
1 Pith paper cite this work, alongside 81 external citations. Polarity classification is still indexing.
1
Pith paper citing it
81
external citations · OpenAlex
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
The SuperActivator Mechanism: Transformers Concentrate Reliable Concept Signals in the Tail
Reliable concept presence in transformers is concentrated in the extreme high-activation tail of in-concept tokens; thresholding that tail improves concept detection and localization.