Truth judgments in LLMs are causally steerable along multiple independent directions forming a cone, not just one axis, across Qwen and Gemma families.
Probing classifiers: Promises, shortcomings, and advances
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
From Directions to Cones: Exploring Multidimensional Representations of Propositional Facts in LLMs
Truth judgments in LLMs are causally steerable along multiple independent directions forming a cone, not just one axis, across Qwen and Gemma families.