finding
active
finding:at-and-af-clusters-show-gradual-reconvergence-in-final-layers-under-threat-template-unlike-bt-and-bf-which-remain-separableaT and aF clusters show gradual reconvergence in final layers under threat template, unlike bT and bF which remain separable
Interpreted as model's internal conflict or moral dilemma during deceptive behavior generation
Source paper
extracted_from(2025) · Kai Wang · Yihao Zhang · Meng Sun
Neighborhood — ranked by edge-count
Claims (1)
claim
- Interpretive claim attributing representational pattern to internal model state during threat-based deception
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.
- Ignition-like effect.
- Statistical grouping of properties based on dependency patterns, enabling deeper understanding of their coherence and interaction.
- Visual geometric evidence for the fundamental entanglement of true/false activations in harder tasks.
- Demonstrates that early-layer probes capture sentence polarity rather than truth.
- Second cluster of the five-cluster grouping, containing properties 11, 4, 6.
- Fig. 7d,e.
- A random head in the same layer shows almost no structure between count and width probesfinding0.725Control for twisting.
- Supporting signature for persona-model collapse: unconditioned moral profiles saturate near ceiling