# Verbalized Sampling: How to Mitigate Mode Collapse and Unlock LLM Diversity

arXiv: https://arxiv.org/abs/2510.01171 · HTML: https://mlsanity.com/paper/2510.01171/

## Key facts

- Cited by 105 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 1 Oct 2025.

## Built on by (most cited first, 30 of 105)

- [Cognitive Alpha Mining via LLM-Driven Code-Based Evolution](https://mlsanity.com/paper/2511.18850/) (2025)
- [When Benign Inputs Lead to Severe Harms: Eliciting Unsafe Unintended Behaviors of Computer-Use Agents](https://mlsanity.com/paper/2602.08235/) (2026)
- [Shepherd: Enabling Programmable Meta-Agents via Reversible Agentic Execution Traces](https://mlsanity.com/paper/2605.10913/) (2026)
- [SPADE: Self-Play in Adaptive Synthetic Executable Environments](https://mlsanity.com/paper/2608.19197/) (2026)
- [The Illusion of Stochasticity in LLMs](https://mlsanity.com/paper/2604.06543/) (2026)
- [CREATE: Testing LLMs for Associative Creativity](https://mlsanity.com/paper/2603.09970/) (2026)
- [AdaPlanBench: Evaluating Adaptive Planning in Large Language Model Agents under World and User Constraints](https://mlsanity.com/paper/2606.05622/) (2026)
- [NOVA: NOise-aware Verbal Confidence CAlibration for Robust Large Language Models in RAG Systems](https://mlsanity.com/paper/2601.11004/) (2026)
- [DiscoverLLM: From Executing Intents to Discovering Them](https://mlsanity.com/paper/2602.03429/) (2026)
- [Reaching Beyond the Mode: RL for Distributional Reasoning in Language Models](https://mlsanity.com/paper/2603.24844/) (2026)
- [LLMs Should Express Uncertainty Explicitly](https://mlsanity.com/paper/2604.05306/) (2026)
- [The Impact of AI-Generated Text on the Internet](https://mlsanity.com/paper/2604.26965/) (2026)
- [Unlocking LLM Creativity in Science through Analogical Reasoning](https://mlsanity.com/paper/2605.11258/) (2026)
- [One Frozen Simulator Is Not Enough: Simulator Collapse in Multi-Agent RL](https://mlsanity.com/paper/2608.12253/) (2026)
- [Verbalizing LLMs' assumptions to explain and control sycophancy](https://mlsanity.com/paper/2604.03058/) (2026)
- [Annotations Mitigate Post-Training Mode Collapse](https://mlsanity.com/paper/2605.09995/) (2026)
- [Reasoning about Intent for Ambiguous Requests](https://mlsanity.com/paper/2511.10453/) (2025)
- [MAR:Multi-Agent Reflexion Improves Reasoning Abilities in LLMs](https://mlsanity.com/paper/2512.20845/) (2025)
- [LLMs Exhibit Significantly Lower Uncertainty in Creative Writing Than Professional Writers](https://mlsanity.com/paper/2602.16162/) (2026)
- [The Alignment Tax: Response Homogenization in Aligned LLMs and Its Implications for Uncertainty Estimation](https://mlsanity.com/paper/2603.24124/) (2026)
- [FlexSQL: Flexible Exploration and Execution Make Better Text-to-SQL Agents](https://mlsanity.com/paper/2605.02815/) (2026)
- [Let the Model Distribute Its Doubt: Confidence Estimation through Verbalized Probability Distribution](https://mlsanity.com/paper/2511.14275/) (2025)
- [PRISM: Pushing the Frontier of Deep Think via Process Reward Model-Guided Inference](https://mlsanity.com/paper/2603.02479/) (2026)
- [TurboEvolve: Towards Fast and Robust LLM-Driven Program Evolution](https://mlsanity.com/paper/2604.18607/) (2026)
- [Breaking the Solver Bottleneck: Training Task Generators at the Learnable Frontier](https://mlsanity.com/paper/2606.18284/) (2026)
- [Dutch Books for Language Models](https://mlsanity.com/paper/2609.02797/) (2026)
- [An Empirical Study on Preference Tuning Generalization and Diversity Under Domain Shift](https://mlsanity.com/paper/2601.05882/) (2026)
- [Decoding as Optimisation on the Probability Simplex: From Top-K to Top-P (Nucleus) to Best-of-K Samplers](https://mlsanity.com/paper/2602.18292/) (2026)
- [PowerFlow: Unlocking the Dual Nature of LLMs via Principled Distribution Matching](https://mlsanity.com/paper/2603.18363/) (2026)
- [Inducing Sustained Creativity and Diversity in Large Language Models](https://mlsanity.com/paper/2603.19519/) (2026)

## Abstract (arXiv)

Post-training alignment often reduces LLM diversity, leading to a phenomenon known as mode collapse. Unlike prior work that attributes this effect to algorithmic limitations, we identify a fundamental, pervasive data-level driver: typicality bias in preference data, whereby annotators systematically favor familiar text as a result of well-established findings in cognitive psychology. We formalize this bias theoretically, verify it on preference datasets empirically, and show that it plays a central role in mode collapse. Motivated by this analysis, we introduce Verbalized Sampling, a simple, training-free prompting strategy to circumvent mode collapse. VS prompts the model to verbalize a probability distribution over a set of responses (e.g., "Generate 5 jokes about coffee and their corresponding probabilities"). Comprehensive experiments show that VS significantly improves performance across creative writing (poems, stories, jokes), dialogue simulation, open-ended QA, and synthetic data generation, without sacrificing factual accuracy and safety. For instance, in creative writing, VS increases diversity by 1.6-2.1x over direct prompting. We further observe an emergent trend that more capable models benefit more from VS. In sum, our work provides a new data-centric perspective on mode collapse and a practical inference-time remedy that helps unlock pre-trained generative diversity.