daDPO: Distribution-Aware DPO for Distilling Conversational Abilities

• URL: http://arxiv.org/abs/2506.15717v1
• Date: Tue, 03 Jun 2025 03:39:29 GMT
• Title: daDPO: Distribution-Aware DPO for Distilling Conversational Abilities
• Authors: Zhengze Zhang, Shiqi Wang, Yiqun Shen, Simin Guo, Dahua Lin, Xiaoliang Wang, Nguyen Cam-Tu, Fei Tan,
• Abstract summary: This paper introduces daDPO (Distribution-Aware DPO), a unified method for preference optimization and distribution-based distillation.<n>We show that daDPO outperforms existing methods in restoring performance for pruned models and enhancing smaller LLM models.
• Score: 48.745922491268004
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large language models (LLMs) have demonstrated exceptional performance across various applications, but their conversational abilities decline sharply as model size decreases, presenting a barrier to their deployment in resource-constrained environments. Knowledge distillation with Direct Preference Optimization (dDPO) has emerged as a promising approach to enhancing the conversational abilities of smaller models using a larger teacher model. However, current methods primarily focus on 'black-box' KD, which only uses the teacher's responses, overlooking the output distribution offered by the teacher. This paper addresses this gap by introducing daDPO (Distribution-Aware DPO), a unified method for preference optimization and distribution-based distillation. We provide rigorous theoretical analysis and empirical validation, showing that daDPO outperforms existing methods in restoring performance for pruned models and enhancing smaller LLM models. Notably, in in-domain evaluation, our method enables a 20% pruned Vicuna1.5-7B to achieve near-teacher performance (-7.3% preference rate compared to that of dDPO's -31%), and allows Qwen2.5-1.5B to occasionally outperform its 7B teacher model (14.0% win rate).

Related papers

• Perception-Aware Policy Optimization for Multimodal Reasoning [79.56070395437898]
  A major source of error in current multimodal reasoning lies in the perception of visual inputs.<n>We propose PAPO, a novel policy gradient algorithm that encourages the model to learn to perceive while learning to reason.<n>We observe a substantial reduction of 30.5% in perception errors, indicating improved perceptual capabilities with PAPO.
  arXiv  Detail & Related papers  (2025-07-08T23:22:34Z)
• Understanding the Performance Gap in Preference Learning: A Dichotomy of RLHF and DPO [51.22869332661607]
  We decompose the performance gap between reinforcement learning from human feedback and direct preference optimization under a representation gap.<n>We show that RLHF, DPO, or online DPO can outperform one another depending on the type of model mis-specifications.
  arXiv  Detail & Related papers  (2025-05-26T09:54:02Z)
• Towards Self-Improvement of Diffusion Models via Group Preference Optimization [10.6096255671291]
  Group Preference Optimization (GPO) is an effective self-improvement method that enhances performance without requiring external data.<n>GPO improves the accurate counting and text rendering capabilities of the Stable Diffusion 3.5 Medium by 20 percentage points.<n>As a plug-and-play method, no extra overhead is introduced during inference.
  arXiv  Detail & Related papers  (2025-05-16T10:04:57Z)
• A Survey of Direct Preference Optimization [103.59317151002693]
  Large Language Models (LLMs) have demonstrated unprecedented generative capabilities.<n>Their alignment with human values remains critical for ensuring helpful and harmless deployments.<n>Direct Preference Optimization (DPO) has recently gained prominence as a streamlined alternative.
  arXiv  Detail & Related papers  (2025-03-12T08:45:15Z)
• Enhancing Knowledge Distillation for LLMs with Response-Priming Prompting [1.9461727843485295]
  We propose a set of novel response-priming prompting strategies to enhance the performance of student models.<n>Our approach fine-tunes a smaller Llama 3.1 8B Instruct model by distilling knowledge from a quantized Llama 3.1 405B Instruct teacher model.<n>We find that Ground Truth prompting results in a 55% performance increase on GSM8K for a distilled Llama 3.1 8B Instruct.
  arXiv  Detail & Related papers  (2024-12-18T20:41:44Z)
• 3D-Properties: Identifying Challenges in DPO and Charting a Path Forward [17.27880657597116]
  We revisit DPO, analyzing its theoretical foundations and empirical performance.<n>We identify three key properties, termed 3D properties, that emerge from DPO's learning process.<n>We propose simple regularization techniques that improve training stability and performance.
  arXiv  Detail & Related papers  (2024-06-11T14:59:24Z)
• MallowsPO: Fine-Tune Your LLM with Preference Dispersions [9.697663437292848]
  Direct Preference Optimization (DPO) has emerged as a popular approach to improve reinforcement learning with human feedback.<n>Inspired by Mallows' theory of preference ranking, we develop in this paper a new approach, the MallowsPO.<n>A distinct feature of this approach is a dispersion index, which reflects the dispersion of human preference to prompts.
  arXiv  Detail & Related papers  (2024-05-23T18:01:11Z)
• D2PO: Discriminator-Guided DPO with Response Evaluation Models [63.71853401569461]
  We propose D2PO, discriminator-guided DPO, for the online setting where preferences are being collected throughout learning. As we collect gold preferences, we use these not only to train our policy, but to train a discriminative response evaluation model to silver-label even more synthetic data for policy training. We show conditions under which silver labeling is most helpful: it is most effective when training the policy with DPO, outperforming traditional PPO, and benefits from maintaining a separate discriminator from the policy model.
  arXiv  Detail & Related papers  (2024-05-02T17:44:41Z)
• Self-Play Preference Optimization for Language Model Alignment [75.83359213697854]
  Recent advancements suggest that directly working with preference probabilities can yield a more accurate reflection of human preferences. We propose a self-play-based method for language model alignment, which treats the problem as a constant-sum two-player game. Our approach, dubbed Self-Play Preference Optimization (SPPO), utilizes iterative policy updates to provably approximate the Nash equilibrium.
  arXiv  Detail & Related papers  (2024-05-01T17:59:20Z)
• Model Extrapolation Expedites Alignment [135.12769233630362]
  We propose a method called ExPO to expedite alignment training with human preferences.<n>We demonstrate that ExPO boosts a DPO model trained with only 20% steps to outperform the fully-trained one.<n>We show that ExPO notably improves existing open-source LLMs on the leading AlpacaEval 2.0 and MT-Bench benchmarks.
  arXiv  Detail & Related papers  (2024-04-25T17:39:50Z)

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.