Euryale 70B (roleplay LoRA on Llama 3.3 70B) scores 1.81, below its base model Llama 3.3 70B at 1.91

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

• Euryale 70B lifts only +1.57 (to 3.38); LoRA fine-tuning capped both default accessibility and latent capacityfinding0.807
  Contrast with Magnum shows LoRA vs full fine-tuning difference in residual headroom
• Suppressing deception/roleplay SAE features in LLaMA 3.3 70B yields 0.96±0.03 consciousness affirmation rate; amplification yields only 0.16±0.05 (z=8.06, p=7.7×10⁻¹⁶)finding0.785
  Core result of Experiment 2: deception feature suppression sharply increases experience claims
• Llama 3.3 70B is the most likely to take on a non-Assistant persona when steered, with even split between human and nonhuman portrayalsfinding0.773
  Model-specific difference in persona susceptibility
• All five judge models consistently rank Llama-3.3-70B as having substantially higher ESR rates than other modelsfinding0.772
  Cross-judge validation of the primary ESR finding across OpenAI, Alibaba, Anthropic, and Google judge models
• LLaMA3.1-70Bconcept0.768
  One of four LLMs selected; larger model with D=8192 embedding dimension; analyzed across proportionally aligned layers.
• LLaMA-3.1-8B: Sbmax = -1.896 ± 0.211, AUSN = -2.119 ± 0.198, peak layer ℓ* = 10 (median)finding0.762
  Seed-pooled geometry-only statistics (per-dev z units).
• Llama-3.3-70B shows multi-attempt rate of 7.4% vs. ≤1.2% for all other models testedfinding0.761
  Supporting finding showing ESR is driven by both higher multi-attempt rates and comparable improvement rates
• Llama-3.3-70B corrected response scores 75/100 rather than 100 due to residual steering effects (Snell's law reference)finding0.760
  Illustrative finding that ESR mitigates but does not fully eliminate steering influence