A dependency graph of 17 values learned from two LLMs predicts side effects of role and SAE steering, but the causal and human-alignment claims are unsupported.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2024 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Are the Values of LLMs Structurally Aligned with Humans? A Causal Perspective
A dependency graph of 17 values learned from two LLMs predicts side effects of role and SAE steering, but the causal and human-alignment claims are unsupported.