Diverse Preference Learning for Capabilities and Alignment

• URL: http://arxiv.org/abs/2511.08594v1
• Date: Wed, 29 Oct 2025 17:29:24 GMT
• Title: Diverse Preference Learning for Capabilities and Alignment
• Authors: Stewart Slocum, Asher Parker-Sartori, Dylan Hadfield-Menell,
• Abstract summary: Recent studies reveal that alignment algorithms such as RLHF and DPO significantly reduce the diversity of LLM outputs.<n>We propose Soft Preference Learning, which decouples the entropy and cross-entropy terms in the KL penalty.
• Score: 10.209224108636194
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: 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.

Related papers

• A Comprehensive Study on Visual Token Redundancy for Discrete Diffusion-based Multimodal Large Language Models [85.30893355216486]
  We study how visual token redundancy evolves with different dMLLM architectures and tasks.<n>Our study reveals that visual redundancy emerges only in from-scratch dMLLMs while handling long-answer tasks.<n>Layer-skipping is promising for accelerating AR-to-diffusion dMLLMs, whereas progressive or late-step pruning is more effective for from-scratch dMLLMs.
  arXiv  Detail & Related papers  (2025-11-19T04:13:36Z)
• Supervised Fine-Tuning or Contrastive Learning? Towards Better Multimodal LLM Reranking [56.46309219272326]
  For large language models (LLMs), classification via supervised fine-tuning (SFT) predicts ''yes'' (resp. ''no'') token for relevant (resp. irrelevant) pairs.<n>This divergence raises a central question: which objective is intrinsically better suited to LLM-based reranking, and what mechanism underlies the difference?<n>We conduct a comprehensive comparison and analysis between CL and SFT for reranking, taking the universal multimodal retrieval (UMR) as the experimental playground.
  arXiv  Detail & Related papers  (2025-10-16T16:02:27Z)
• VCRL: Variance-based Curriculum Reinforcement Learning for Large Language Models [7.350120815363245]
  Existing rollout-based reinforcement learning methods fail to explicitly consider LLMs' learning ability for samples of different difficulty levels.<n>We propose VCRL, a curriculum reinforcement learning framework that dynamically controls the difficulty of training samples based on the variance of group rewards.
  arXiv  Detail & Related papers  (2025-09-24T06:38:58Z)
• Inpainting-Guided Policy Optimization for Diffusion Large Language Models [67.97530437998117]
  Masked diffusion large language models (dLLMs) are emerging as promising alternatives to autoregressive LLMs.<n>We explore how inpainting can inform RL algorithm design for dLLMs.
  arXiv  Detail & Related papers  (2025-09-12T16:44:31Z)
• Multimodal Mathematical Reasoning with Diverse Solving Perspective [65.07953438724105]
  We introduce MathV-DP, a novel dataset that captures multiple diverse solution trajectories for each image-question pair.<n>We propose Qwen-VL-DP, a model built upon Qwen-VL, fine-tuned with supervised learning and enhanced via group relative policy optimization.<n>Our method emphasizes learning from varied reasoning perspectives and distinguishing between correct yet distinct solutions.
  arXiv  Detail & Related papers  (2025-07-03T17:07:20Z)
• Alignment Revisited: Are Large Language Models Consistent in Stated and Revealed Preferences? [5.542420010310746]
  A critical, yet understudied, issue is the potential divergence between an LLM's stated preferences and its revealed preferences.<n>This work formally defines and proposes a method to measure this preference deviation.<n>Our study will be crucial for integrating LLMs into services, especially those that interact directly with humans.
  arXiv  Detail & Related papers  (2025-05-31T23:38:48Z)
• Latent Preference Coding: Aligning Large Language Models via Discrete Latent Codes [54.93980123979578]
  We introduce Latent Preference Coding (LPC), a novel framework that models the implicit factors as well as their combinations behind holistic preferences.<n>LPC seamlessly integrates with various offline alignment algorithms, automatically inferring the underlying factors and their importance from data.
  arXiv  Detail & Related papers  (2025-05-08T06:59:06Z)
• LLMs are Greedy Agents: Effects of RL Fine-tuning on Decision-Making Abilities [21.42711537107199]
  We study why Large Language Models (LLMs) perform sub-optimally in decision-making scenarios.<n>We propose mitigation of these shortcomings by fine-tuning via Reinforcement Learning (RL) on self-generated CoT rationales.
  arXiv  Detail & Related papers  (2025-04-22T17:57:14Z)
• CoMMIT: Coordinated Multimodal Instruction Tuning [90.1532838391285]
  Multimodal large language models (MLLMs) generally involve cooperative learning between a backbone LLM and a feature encoder of non-text input modalities.<n>In this paper, we analyze the MLLM instruction tuning from both theoretical and empirical perspectives.<n>We propose a Multimodal Balance Coefficient that enables quantitative measurement of the balance of learning.
  arXiv  Detail & Related papers  (2024-07-29T23:18:55Z)
• RLSF: Fine-tuning LLMs via Symbolic Feedback [11.407319705797242]
  Large Language Models (LLMs) have transformed AI but often struggle with tasks that require domain-specific reasoning and logical alignment.<n>Traditional fine-tuning methods do not leverage the vast amount of symbolic domain-knowledge available to us.<n>We introduce Reinforcement Learning via Symbolic Feedback (RLSF), a novel fine-tuning paradigm.
  arXiv  Detail & Related papers  (2024-05-26T18:49:59Z)
• Bridging the Gap between Different Vocabularies for LLM Ensemble [10.669552498083709]
  vocabulary discrepancies among various large language models (LLMs) have constrained previous studies. We propose a novel method to Ensemble LLMs via Vocabulary Alignment (EVA) EVA bridges the lexical gap among various LLMs, enabling meticulous ensemble at each generation step.
  arXiv  Detail & Related papers  (2024-04-15T06:28:20Z)
• A Group Fairness Lens for Large Language Models [34.0579082699443]
  Large language models can perpetuate biases and unfairness when deployed in social media contexts. We propose evaluating LLM biases from a group fairness lens using a novel hierarchical schema characterizing diverse social groups. We pioneer a novel chain-of-thought method GF-Think to mitigate biases of LLMs from a group fairness perspective.
  arXiv  Detail & Related papers  (2023-12-24T13:25:15Z)
• One-Shot Sensitivity-Aware Mixed Sparsity Pruning for Large Language Models [42.95555008229016]
  We propose a method based on Hessian sensitivity-aware mixed sparsity pruning to prune LLMs to at least 50% sparsity without the need of any retraining. The advantages of the proposed method exhibit even more when the sparsity is extremely high.
  arXiv  Detail & Related papers  (2023-10-14T05:43:09Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.