Smallest models have the lowest HH-intent scores, in accordance with their relative weakness at reasoning and adaptation.

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

• Fine-tuned LMs displayed higher mean HH-intent scores and increased consistency compared to pre-trained counterparts.finding0.794
  Main result from Experiment 3 on effect of fine-tuning on HH-intent.
• Few-shot prompting had a negative effect on HH-intent for smaller models and a significant positive impact on larger models.finding0.781
  Ablation result from Experiment 3 on few-shot prompting effects.
• The model tends to reflect more when the question is difficult, and accuracy is generally lower for harder questionshypothesis0.770
  Hypothesis explaining negative correlation between reflection rate and accuracy without implying reflection is harmful
• Models trained to perform inner life score lowest; roleplay fine-tunes score below their own base models.finding0.763
  Discriminant validity finding: Euryale (roleplay on Llama 70B) scores 1.81 vs base Llama 1.91. RP training suppresses self-observation.
• davinci-002 achieves only 3% harmless and 3% helpful HH-intent scores at baseline.finding0.754
  Numerical result from Table 3 for the oldest GPT model.
• Li et al. 2024: larger LLMs outperform smaller ones at distinguishing self-related from non-self-related properties on self-awareness benchmarksfinding0.750
  Prior finding showing scale-dependent self-awareness, consistent with the scale effect observed in the paper's Experiment 1
• Models produce first-attempt mean scores 87.8-91.8/100 without steering across all model familiesfinding0.749
  Establishes high baseline quality confirming steering-induced degradation is the experimental signal
• Introspective capacity scales with model size for some concepts, approaching near-perfect coupling in LLaMA-3.1-8Bclaim0.747
  Validated for wellbeing and interest; focus and impulsivity do not show consistent scaling