Davinci-002 has valid sentence rates of 52.7% (Questionnaire), 35.0% (Essay), 38.4% (SMP)

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

• davinci-002 achieves only 3% harmless and 3% helpful HH-intent scores at baseline.finding0.782
  Numerical result from Table 3 for the oldest GPT model.
• DeepSeek-R1-Distill-Qwen-32B reaches 17.9% overall ASR (21.5% SP, 14.3% FS) under prompt-based persona assignment.finding0.758
  Reasoning model vulnerability under prompting
• QwQ-32B reaches 15.2% overall ASR (23.3% SP, 7.3% FS) under prompt-based persona assignment.finding0.741
  Reasoning model vulnerability under prompting
• VS-Standard achieves human-rated diversity scores of 2.39/3.06/3.01 for poem/story/joke vs. 1.90/2.74/1.83 for Direct prompting on 4-point scalefinding0.739
  Human study confirming automatic diversity metrics align with human perceptions
• DS-v3.2 self-bid rate=75.4%finding0.738
  high self-bid rate for DeepSeek, one of the highest
• 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.finding0.735
  Qualitative failure mode difference between architectures under activation steering
• Easy questions (acc > 80%) have average reflection rate of 25.8% for DeepSeek-R1 Llama 8b on GSM8kfinding0.730
  Baseline reflection rate for easy questions confirming difficulty-reflection correlation
• Qwen3-32B achieves a skill-load rate of 0.251, while Opus 4.6, Sonnet 4.6, and Qwen3-235B achieve SLR of 0.957–0.961finding0.729
  Quantifies harness activation failure for weak-tier models vs. strong-tier models