VS-Standard achieves KL divergence of 0.027 from uniform distribution on dice roll simulation, vs. 0.926 for Direct prompting

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

• VS-Standard achieves KL divergence of 0.12-0.13 from pretraining distribution on US state naming task, vs. 14.89-16.16 for Direct promptingfinding0.890
  Strong empirical evidence that VS recovers pretraining distribution while direct prompting collapses
• VS-Standard achieves KL divergence of 0.54 from pretraining distribution on Open-Ended QA, vs. 3.14 for Direct and 0.58 for Sequencefinding0.874
  Shows VS substantially better approximates the pretraining distribution than baseline methods
• 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 scalefinding0.801
  Human study confirming automatic diversity metrics align with human perceptions
• VS-Multi achieves average accuracy of 37.5% on math benchmarks (avg of MATH500, OlympiadBench, Minerva) with Gemini-2.5-Flash as generator and Qwen3-4B as SFT model, vs. 40.7% baselinefinding0.772
  Best VS result in synthetic data generation for math, demonstrating downstream improvement through diversity
• VS-Standard achieves 1.86x diversity gain at only 1.12x cost increase over Direct prompting on poem generationfinding0.766
  Cost-diversity trade-off analysis showing VS's practical efficiency
• Representational divergence (as measured by EMD) can predict lower out-of-distribution intervention performanceclaim0.760
  Practical utility of reducing divergence demonstrated through regression analysis
• VS recovers 66.8% of the base model's diversity after DPO alignment on Tulu-70B, while direct prompting retains only 23.8%finding0.754
  Quantifies how much of the base model's diversity VS can recover compared to baseline prompting
• Clear accuracy stratification across three reflection levels on cruxeval_o_adv: Triggered (.065/.247) > Intrinsic (.040/.133) > No Reflection (.017/.051) for Qwen2.5-3B/Gemma3-4B-ITfinding0.749
  Core empirical result validating the three-level reflection framework on code reasoning.