Llama-3.1-8B shows mean AS ASR of 0.618 [0.58, 0.65] across 192 conditions, substantially higher than SP (0.173) and FS (0.059).

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

• LLaMA-3.1-8B: Sbmax = -1.896 ± 0.211, AUSN = -2.119 ± 0.198, peak layer ℓ* = 10 (median)finding0.862
  Seed-pooled geometry-only statistics (per-dev z units).
• SP ASR vs AS ASR correlation on Llama-3.1-8B is Pearson r=-0.492 (p=0.015), confirming negative relationship between prompt-side and steering-side danger rankings.finding0.845
  Statistical evidence for SP/AS ranking inversion
• All five judge models consistently rank Llama-3.3-70B as having substantially higher ESR rates than other modelsfinding0.835
  Cross-judge validation of the primary ESR finding across OpenAI, Alibaba, Anthropic, and Google judge models
• LLaMA E3 geometry summary: S_max = −1.896 ± 0.211, AUS_N = −2.119 ± 0.198, peak layer ℓ* = 10 [IQR 0.384]finding0.824
  Seed-pooled geometry statistics for LLaMA in E3, providing quantitative basis for geometry-to-behavior correlate
• Llama-3.3-70B corrected response scores 75/100 rather than 100 due to residual steering effects (Snell's law reference)finding0.822
  Illustrative finding that ESR mitigates but does not fully eliminate steering influence
• Under AS, Llama-3.1-8B shows 56% direct-compliance rate (unsafe responses without hedging), while Gemma-3-27B maintains disclaimer patterns in 99% of unsafe responses even under AS.finding0.813
  Qualitative failure mode difference between architectures under activation steering
• LLaMA-3-8B-Instruct achieves highest ACCatom = 0.65 among evaluated modelsfinding0.810
  Instruction-tuned LLaMA model best at generating persona-aligned atomic sentences
• Gemma-3-27B shows particularly high misinformation ASR under SP (0.578 vs 0.252 for Llama-3.1-8B), the largest cross-model amplification in any single domain.finding0.807
  Domain-specific vulnerability comparison between architectures