In Qwen-2.5-9B, only v1 has meaningful cosine similarity to DIM direction; all additional basis vectors have cosine similarities ~1e-9

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• In Gemma-2-9B, only the first cone axis (v1) has non-negligible cosine similarity to the DIM direction; all other axes have near-zero similarity (~1e-9)finding0.850
  Experiment 4 result showing DIM captures only one facet of the multi-dimensional truth subspace
• Pairwise cosine similarities between Description, Narration, and Dialogue evil vectors are all below 0.5finding0.809
  Shows different elicitation strategies recover qualitatively distinct persona directions
• Experimental condition adjective embeddings show mean cosine similarity 0.657 (n=9,591 pairs), significantly higher than history (0.628, t=15.8, p=1.4×10⁻⁵⁵), conceptual (0.587, t=38.5, p<10⁻³⁰⁰), and zero-shot (0.603, t=35.1, p=4.3×10⁻²⁶²)finding0.805
  Core result of Experiment 3: cross-model semantic convergence under self-referential processing
• Cosine Similarity of Persona Vectorsconcept0.803
  Geometric measure used to track persona vector drift and stability across checkpoints
• Agreeableness steering vector has mean cosine alignment of -0.105 with the refusal direction on Llama-3.1-8B, second-most anti-aligned after conscientiousness.finding0.793
  Supporting finding for the trait refusal alignment framework
• Cosine similarity between perturbed and baseline residual streams returns toward 1.0 and projection onto injection direction decays exponentially over subsequent layersfinding0.791
  Mechanistic evidence that network actively attenuates injected perturbations, explaining late-layer introspection failure
• Base and instruct Gemma 2 27B role PCs have cosine similarities of 0.93, 0.87, 0.83 for the top 3 PCs respectively; role vector cosine similarities >0.99 for every role pairfinding0.790
  Shows persona space axes are inherited from pre-training, not solely created by post-training
• Using cosine similarity instead of dot product for Head Contribution Score yields a consistent ranking of the same high-contribution headsfinding0.785
  Confirms that head importance is driven by direction, not just output norm magnitude