finding
active
finding:most-independent-dimension-pair-is-aesthetic-response-and-boundary-awareness-rho-0-553-most-correlated-is-prediction-error-and-conceptual-crystallization-rho-0-886Most independent dimension pair is aesthetic_response and boundary_awareness (rho=0.553); most correlated is prediction_error and conceptual_crystallization (rho=0.886)
Characterizes internal structure of the six scoring dimensions
Source paper
extracted_from(2026) · Borzov, Anton
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Constitutional AI fingerprint in dimension profile; training that makes models self-observant also makes them polished at cost to aliveness
- Strong scaling trend for introspective fidelity when excluding invalid steering-sign pairs
- Interpretive finding from dimension profile analysis: training for honest limits comes at cost to aliveness.
- SAE features are not simply mirroring individual neurons.
- Statistical grouping of properties based on dependency patterns, enabling deeper understanding of their coherence and interaction.
- Validates robustness of alignment metric choice
- Heavy alignment includes both CAI (low lift) and heavy-RLHF (high lift); predictor is alignment type not depth