VS-Standard achieves 1.86x diversity gain at only 1.12x cost increase over Direct prompting on poem generation

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

• 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 scalefinding0.856
  Human study confirming automatic diversity metrics align with human perceptions
• VS increases diversity by 1.6-2.1x over direct prompting on creative writing tasks (poem, story, joke)finding0.849
  Core empirical result demonstrating VS's effectiveness on creative writing diversity
• After DPO stage, VS outperforms direct prompting by 182.6% on diversity in poem continuation using Tulu-70Bfinding0.787
  Demonstrates the magnitude of VS's advantage over direct prompting after aggressive alignment training
• 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 promptingfinding0.786
  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 Sequencefinding0.773
  Shows VS substantially better approximates the pretraining distribution than baseline methods
• VS-Standard achieves KL divergence of 0.027 from uniform distribution on dice roll simulation, vs. 0.926 for Direct promptingfinding0.766
  Shows VS enables LLMs to better approximate random behavior compared to direct prompting
• Diversity Threshold Generation results in average 137% increase in NLI Diversity compared to standard generationfinding0.758
  Key headline result of the DTG procedure across all conditions
• VS improves human evaluation scores by 25.7% on creative writing compared to direct promptingfinding0.757
  Human study result validating automatic diversity metrics for creative writing tasks