mlsanityData to 7 Oct 2026

Trading Off Diversity and Quality in Natural Language Generation

arXiv 2004.10450 ↗First posted 22 Apr 2020cs.CLcited by 94

Not on the map. Why?

Abstract (arXiv)

For open-ended language generation tasks such as storytelling and dialogue, choosing the right decoding algorithm is critical to controlling the tradeoff between generation quality and diversity. However, there presently exists no consensus on which decoding procedure is best or even the criteria by which to compare them. We address these issues by casting decoding as a multi-objective optimization problem aiming to simultaneously maximize both response quality and diversity. Our framework enables us to perform the first large-scale evaluation of decoding methods along the entire quality-diversity spectrum. We find that when diversity is a priority, all methods perform similarly, but when quality is viewed as more important, the recently proposed nucleus sampling (Holtzman et al. 2019) outperforms all other evaluated decoding algorithms. Our experiments also confirm the existence of the `likelihood trap', the counter-intuitive observation that high likelihood sequences are often surprisingly low quality. We leverage our findings to create and evaluate an algorithm called selective sampling which tractably approximates globally-normalized temperature sampling.

Key facts. Cited by 94 papers in the map (data through 7 Oct 2026). First posted on arXiv 22 Apr 2020.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2020: 1; 2021: 5; 2022: 19; 2023: 17; 2024: 18; 2025: 26; 2026: 8.

94 papers.

Sort:

By area:

  1. 2020Learning to summarize from human feedbackcited by 2,562
  2. 2023A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questionscited by 1,582
  3. 2021MAUVE: Measuring the Gap Between Neural Text and Human Text using Divergence Frontierscited by 466
  4. 2022DiffuSeq: Sequence to Sequence Text Generation with Diffusion Modelscited by 439
  5. 2022SSD-LM: Semi-autoregressive Simplex-based Diffusion Language Model for Text Generation and Modular Controlcited by 191
  6. 2025Scalable Best-of-N Selection for Large Language Models via Self-Certaintycited by 182
  7. 2024From Decoding to Meta-Generation: Inference-time Algorithms for Large Language Modelscited by 136
  8. 2022Locally Typical Samplingcited by 108
  9. 2023The Curious Decline of Linguistic Diversity: Training Language Models on Synthetic Textcited by 97
  10. 2024Is Temperature the Creativity Parameter of Large Language Models?cited by 92
  11. 2022On Decoding Strategies for Neural Text Generatorscited by 89
  12. 2023Effective Prompt Extraction from Language Modelscited by 83
38 more
  1. 2025Jointly Reinforcing Diversity and Quality in Language Model Generationscited by 79
  2. 2022Do Language Models Plagiarize?cited by 72
  3. 2022RankGen: Improving Text Generation with Large Ranking Modelscited by 72
  4. 2022Real or Fake Text?: Investigating Human Ability to Detect Boundaries Between Human-Written and Machine-Generated Textcited by 63
  5. 2025Evaluating the Diversity and Quality of LLM Generated Contentcited by 51
  6. 2024Forcing Diffuse Distributions out of Language Modelscited by 48
  7. 2025Diverse Preference Optimizationcited by 48
  8. 2023KCTS: Knowledge-Constrained Tree Search Decoding with Token-Level Hallucination Detectioncited by 43
  9. 2025Modifying Large Language Model Post-Training for Diverse Creative Writingcited by 39
  10. 2024Benchmarking Linguistic Diversity of Large Language Modelscited by 37
  11. 2025The Homogenizing Effect of Large Language Models on Human Expression and Thoughtcited by 35
  12. 2023Probing the Creativity of Large Language Models: Can models produce divergent semantic association?cited by 31
  13. 2022A Well-Composed Text is Half Done! Composition Sampling for Diverse Conditional Generationcited by 28
  14. 2024Mitigating Hallucinations in Large Vision-Language Models (LVLMs) via Language-Contrastive Decoding (LCD)cited by 23
  15. 2024Collective Critics for Creative Story Generationcited by 21
  16. 2023On the Efficacy of Sampling Adapterscited by 17
  17. 2023Vision-Language Instruction Tuning: A Review and Analysiscited by 17
  18. 2022On the probability-quality paradox in language generationcited by 16
  19. 2023Diffusion Models in NLP: A Surveycited by 16
  20. 2023A Survey of Diffusion Models in Natural Language Processingcited by 16
  21. 2023MixCE: Training Autoregressive Language Models by Mixing Forward and Reverse Cross-Entropiescited by 16
  22. 2025Control the Temperature: Selective Sampling for Diverse and High-Quality LLM Outputscited by 15
  23. 2023Affective and Dynamic Beam Search for Story Generationcited by 12
  24. 2024Measuring memorization in RLHF for code completioncited by 12
  25. 2024Generating Diverse and High-Quality Texts by Minimum Bayes Risk Decodingcited by 11
  26. 2023AutoHall: Automated Factuality Hallucination Dataset Generation for Large Language Modelscited by 10
  27. 2024Diversity-Rewarded CFG Distillationcited by 9
  28. 2024Growing a Tail: Increasing Output Diversity in Large Language Modelscited by 9
  29. 2024Adaptive Decoding via Latent Preference Optimizationcited by 9
  30. 2025Multi-LLM Collaborative Search for Complex Problem Solvingcited by 9
  31. 2022Can Question Rewriting Help Conversational Question Answering?cited by 8
  32. 2023Language Model Decoding as Direct Metrics Optimizationcited by 8
  33. 2022An Analysis of the Effects of Decoding Algorithms on Fairness in Open-Ended Language Generationcited by 7
  34. 2022Best-k Search Algorithm for Neural Text Generationcited by 6
  35. 2023Diverse and Faithful Knowledge-Grounded Dialogue Generation via Sequential Posterior Inferencecited by 6
  36. 2022Language Model Decoding as Likelihood-Utility Alignmentcited by 5
  37. 2023Hierarchical Evaluation Framework: Best Practices for Human Evaluationcited by 5
  38. 2025Emotional Supporters often Use Multiple Strategies in a Single Turncited by 5

50 most cited of 94 All 94 (JSON)