Positive-Unlabeled Reinforcement Learning Distillation for On-Premise Small Models

• URL: http://arxiv.org/abs/2601.20687v1
• Date: Wed, 28 Jan 2026 15:14:50 GMT
• Title: Positive-Unlabeled Reinforcement Learning Distillation for On-Premise Small Models
• Authors: Zhiqiang Kou, Junyang Chen, Xin-Qiang Cai, Xiaobo Xia, Ming-Kun Xie, Dong-Dong Wu, Biao Liu, Yuheng Jia, Xin Geng, Masashi Sugiyama, Tat-Seng Chua,
• Abstract summary: We propose a positive-unlabeled (PU) reinforcement learning distillation method for on-premise small-model deployment.<n>Our method distills the teacher's preference-optimization capability from black-box generations into a locally trainable student.<n>Experiments demonstrate that our method achieves consistently strong performance under a low-cost setting.
• Score: 130.8912476550625
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Due to constraints on privacy, cost, and latency, on-premise deployment of small models is increasingly common. However, most practical pipelines stop at supervised fine-tuning (SFT) and fail to reach the reinforcement learning (RL) alignment stage. The main reason is that RL alignment typically requires either expensive human preference annotation or heavy reliance on high-quality reward models with large-scale API usage and ongoing engineering maintenance, both of which are ill-suited to on-premise settings. To bridge this gap, we propose a positive-unlabeled (PU) RL distillation method for on-premise small-model deployment. Without human-labeled preferences or a reward model, our method distills the teacher's preference-optimization capability from black-box generations into a locally trainable student. For each prompt, we query the teacher once to obtain an anchor response, locally sample multiple student candidates, and perform anchor-conditioned self-ranking to induce pairwise or listwise preferences, enabling a fully local training loop via direct preference optimization or group relative policy optimization. Theoretical analysis justifies that the induced preference signal by our method is order-consistent and concentrates on near-optimal candidates, supporting its stability for preference optimization. Experiments demonstrate that our method achieves consistently strong performance under a low-cost setting.

Related papers

• Multi-Preference Lambda-weighted Listwise DPO for Small-Scale Model Alignment [5.276657230880984]
  Large language models (LLMs) demonstrate strong generalization across a wide range of language tasks, but often generate outputs that misalign with human preferences.<n>Direct Optimization Preference (DPO) simplifies the process by treating alignment as a classification task over binary preference pairs.<n>We propose Multi-Preference Lambda-weighted Listwise DPO, which allows the model to learn from more detailed human feedback.<n>Our method consistently outperforms standard DPO on alignment while enabling efficient, controllable, and fine-grained adaptation suitable for real-world deployment.
  arXiv  Detail & Related papers  (2025-06-24T16:47:17Z)
• Latent Embedding Adaptation for Human Preference Alignment in Diffusion Planners [16.863492060519157]
  This work addresses the challenge of personalizing trajectories generated in automated decision-making systems.<n>We propose a resource-efficient approach that enables rapid adaptation to individual users' preferences.
  arXiv  Detail & Related papers  (2025-03-24T05:11:58Z)
• Prompt Tuning with Diffusion for Few-Shot Pre-trained Policy Generalization [55.14484317645865]
  We develop a conditional diffusion model to produce exceptional quality prompts for offline reinforcement learning tasks. We show that the Prompt diffuser is a robust and effective tool for the prompt-tuning process, demonstrating strong performance in the meta-RL tasks.
  arXiv  Detail & Related papers  (2024-11-02T07:38:02Z)
• Self-supervised Preference Optimization: Enhance Your Language Model with Preference Degree Awareness [27.43137305486112]
  We propose a novel Self-supervised Preference Optimization (SPO) framework, which constructs a self-supervised preference degree loss combined with the alignment loss. The results demonstrate that SPO can be seamlessly integrated with existing preference optimization methods to achieve state-of-the-art performance.
  arXiv  Detail & Related papers  (2024-09-26T12:37:26Z)
• Preference Learning Algorithms Do Not Learn Preference Rankings [62.335733662381884]
  We study the conventional wisdom that preference learning trains models to assign higher likelihoods to more preferred outputs than less preferred outputs. We find that most state-of-the-art preference-tuned models achieve a ranking accuracy of less than 60% on common preference datasets.
  arXiv  Detail & Related papers  (2024-05-29T21:29:44Z)
• Robust Preference Optimization through Reward Model Distillation [68.65844394615702]
  Direct Preference Optimization (DPO) is a popular offline alignment method that trains a policy directly on preference data.<n>We analyze this phenomenon and use distillation to get a better proxy for the true preference distribution over generation pairs.<n>Our results show that distilling from such a family of reward models leads to improved robustness to distribution shift in preference annotations.
  arXiv  Detail & Related papers  (2024-05-29T17:39:48Z)
• Direct Preference Optimization: Your Language Model is Secretly a Reward Model [119.65409513119963]
  We introduce a new parameterization of the reward model in RLHF that enables extraction of the corresponding optimal policy in closed form. The resulting algorithm, which we call Direct Preference Optimization (DPO), is stable, performant, and computationally lightweight. Our experiments show that DPO can fine-tune LMs to align with human preferences as well as or better than existing methods.
  arXiv  Detail & Related papers  (2023-05-29T17:57:46Z)
• Reinforcement Learning from Diverse Human Preferences [68.4294547285359]
  This paper develops a method for crowd-sourcing preference labels and learning from diverse human preferences. The proposed method is tested on a variety of tasks in DMcontrol and Meta-world. It has shown consistent and significant improvements over existing preference-based RL algorithms when learning from diverse feedback.
  arXiv  Detail & Related papers  (2023-01-27T15:18:54Z)

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.