Steering vectors are unreliable when the target behavior does not correspond to a coherent, well-separated linear direction in activation space, and this coherence can be measured from training data.
SIMSUM : Document-level text simplification via simultaneous summarization
1 Pith paper cite this work, alongside 11 external citations. Polarity classification is still indexing.
1
Pith paper citing it
11
external citations · OpenAlex
citation-role summary
other 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
other 1polarities
unclear 1representative citing papers
citing papers explorer
-
Understanding (Un)Reliability of Steering Vectors in Language Models
Steering vectors are unreliable when the target behavior does not correspond to a coherent, well-separated linear direction in activation space, and this coherence can be measured from training data.