# Diverse Preference Learning for Capabilities and Alignment

arXiv: https://arxiv.org/abs/2511.08594 · HTML: https://mlsanity.com/paper/2511.08594/

## Key facts

- Cited by 25 papers in the map (data through 14 Oct 2026).
- Placed in LLM preference optimization methods, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 38 of its 51 references that are in the map.
- First posted on arXiv 29 Oct 2025.

## Placed in

- LLM alignment / RLHF / preference optimisation / [LLM preference optimization methods](https://mlsanity.com/alignment/llm-preference-optimization-methods/)

## Builds on (most cited first, 30 of 38)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://mlsanity.com/paper/1908.10084/) (2019)
- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Measuring Mathematical Problem Solving With the MATH Dataset](https://mlsanity.com/paper/2103.03874/) (2021)
- [Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor](https://mlsanity.com/paper/1801.01290/) (2018)
- [Deep reinforcement learning from human preferences](https://mlsanity.com/paper/1706.03741/) (2017)
- [Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2204.05862/) (2022)
- [Tree of Thoughts: Deliberate Problem Solving with Large Language Models](https://mlsanity.com/paper/2305.10601/) (2023)
- [TruthfulQA: Measuring How Models Mimic Human Falsehoods](https://mlsanity.com/paper/2109.07958/) (2021)
- [Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters](https://mlsanity.com/paper/2408.03314/) (2024)
- [Language Models (Mostly) Know What They Know](https://mlsanity.com/paper/2207.05221/) (2022)
- [Enhancing Chat Language Models by Scaling High-quality Instructional Conversations](https://mlsanity.com/paper/2305.14233/) (2023)
- [Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedback](https://mlsanity.com/paper/2305.14975/) (2023)
- [Zephyr: Direct Distillation of LM Alignment](https://mlsanity.com/paper/2310.16944/) (2023)
- [UltraFeedback: Boosting Language Models with Scaled AI Feedback](https://mlsanity.com/paper/2310.01377/) (2023)
- [Understanding the Effects of RLHF on LLM Generalisation and Diversity](https://mlsanity.com/paper/2310.06452/) (2023)
- [Nash Learning from Human Feedback](https://mlsanity.com/paper/2312.00886/) (2023)
- [Does Writing with Language Models Reduce Content Diversity?](https://mlsanity.com/paper/2309.05196/) (2023)
- [Maximum Entropy RL (Provably) Solves Some Robust RL Problems](https://mlsanity.com/paper/2103.06257/) (2021)
- [Beyond Reverse KL: Generalizing Direct Preference Optimization with Diverse Divergence Constraints](https://mlsanity.com/paper/2309.16240/) (2023)
- [Political Compass or Spinning Arrow? Towards More Meaningful Evaluations for Values and Opinions in Large Language Models](https://mlsanity.com/paper/2402.16786/) (2024)
- [Turning Up the Heat: Min-p Sampling for Creative and Coherent LLM Outputs](https://mlsanity.com/paper/2407.01082/) (2024)
- [A Minimaximalist Approach to Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2401.04056/) (2024)
- [Distributional Preference Learning: Understanding and Accounting for Hidden Context in RLHF](https://mlsanity.com/paper/2312.08358/) (2023)
- [MaxMin-RLHF: Alignment with Diverse Human Preferences](https://mlsanity.com/paper/2402.08925/) (2024)
- [Back to Basics: Revisiting REINFORCE Style Optimization for Learning from Human Feedback in LLMs](https://mlsanity.com/paper/2402.14740/) (2024)

## Built on by (most cited first, 25 of 25)

- [Where does output diversity collapse in post-training?](https://mlsanity.com/paper/2604.16027/) (2026)
- [Group-Aware Reinforcement Learning for Output Diversity in Large Language Models](https://mlsanity.com/paper/2511.12596/) (2025)
- [DPWriter: Reinforcement Learning with Diverse Planning Branching for Creative Writing](https://mlsanity.com/paper/2601.09609/) (2026)
- [PathWise: Planning through World Model for Automated Heuristic Design via Self-Evolving LLMs](https://mlsanity.com/paper/2601.20539/) (2026)
- [Expected Return Causes Outcome-Level Mode Collapse in Reinforcement Learning and How to Fix It with Inverse Probability Scaling](https://mlsanity.com/paper/2601.21669/) (2026)
- [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)
- [Love First, Know Later: Persona-Based Romantic Compatibility Through LLM Text World Engines](https://mlsanity.com/paper/2512.11844/) (2025)
- [Time-Annealed Perturbation Sampling: Diverse Generation for Diffusion Language Models](https://mlsanity.com/paper/2601.22629/) (2026)
- [PERSPECTRA: A Scalable and Configurable Pluralist Benchmark of Perspectives from Arguments](https://mlsanity.com/paper/2602.08716/) (2026)
- [Exploring Diverse Generation Paths via Inference-time Stiefel Activation Steering](https://mlsanity.com/paper/2601.22010/) (2026)
- [Detecting RLVR Training Data via Structural Convergence of Reasoning](https://mlsanity.com/paper/2602.11792/) (2026)
- [AI Alignment From Social Choice Perspectives](https://mlsanity.com/paper/2606.21550/) (2026)
- [Minority-Aware Satisfaction Estimation in Dialogue Systems via Preference-Adaptive Reinforcement Learning](https://mlsanity.com/paper/2511.05407/) (2025)
- [VISPA: Pluralistic Alignment via Automatic Value Selection and Activation](https://mlsanity.com/paper/2601.12758/) (2026)
- [Quality-constrained Entropy Maximization Policy Optimization for LLM Diversity](https://mlsanity.com/paper/2602.15894/) (2026)
- [Forty Shades of Blue: Quality-Diversity Alignment via Mode-Conditioned Reinforcement Learning](https://mlsanity.com/paper/2609.14896/) (2026)
- [Know Thy Reasoner: Not All Language Models Explore Alike](https://mlsanity.com/paper/2604.10827/) (2026)
- [Entropy Centroids as Intrinsic Rewards for Test-Time Scaling](https://mlsanity.com/paper/2604.26173/) (2026)
- [GENIE: A Fine-Grained Measure for Novelty](https://mlsanity.com/paper/2606.12790/) (2026)
- [Collective cooperation without individual fidelity in LLM agents](https://mlsanity.com/paper/2606.30454/) (2026)
- [When Does the Best Sampling Temperature Rise with the Budget? Sufficient Conditions for Pass@k](https://mlsanity.com/paper/2608.14665/) (2026)
- [Direct Diversity Optimization for Diverse Successful Trajectories in Preference Post-Training](https://mlsanity.com/paper/2609.10052/) (2026)
- [Fine-Tuning Fixes Mode Collapse and Over-Dispersion in LLMs](https://mlsanity.com/paper/2609.16454/) (2026)
- [TRACE: Trajectory Return Attribution and Contrastive Erasure for Multi-Turn Safety](https://mlsanity.com/paper/2610.01323/) (2026)

## Abstract (arXiv)

The ability of LLMs to represent diverse perspectives is critical as they increasingly impact society. However, recent studies reveal that alignment algorithms such as RLHF and DPO significantly reduce the diversity of LLM outputs. Not only do aligned LLMs generate text with repetitive structure and word choice, they also approach problems in more uniform ways, and their responses reflect a narrower range of societal perspectives. We attribute this problem to the KL divergence regularizer employed in preference learning algorithms. This causes the model to systematically overweight majority opinions and sacrifice diversity in its outputs. To address this, we propose Soft Preference Learning, which decouples the entropy and cross-entropy terms in the KL penalty - allowing for fine-grained control over LLM generation diversity. From a capabilities perspective, LLMs trained using Soft Preference Learning attain higher accuracy on difficult repeated sampling tasks and produce outputs with greater semantic and lexical diversity. From an alignment perspective, they are capable of representing a wider range of societal viewpoints and display improved logit calibration. Notably, Soft Preference Learning resembles, but is a Pareto improvement over, standard temperature scaling.