Mix Data or Merge Models? Balancing the Helpfulness, Honesty, and Harmlessness of Large Language Model via Model Merging

• URL: http://arxiv.org/abs/2502.06876v2
• Date: Thu, 13 Feb 2025 06:28:33 GMT
• Title: Mix Data or Merge Models? Balancing the Helpfulness, Honesty, and Harmlessness of Large Language Model via Model Merging
• Authors: Jinluan Yang, Dingnan Jin, Anke Tang, Li Shen, Didi Zhu, Zhengyu Chen, Daixin Wang, Qing Cui, Zhiqiang Zhang, Jun Zhou, Fei Wu, Kun Kuang,
• Abstract summary: This paper establishes the first comprehensive benchmark for model merging in large language models (LLMs) Our analysis reveals three pivotal insights: (i) previously overlooked collaborative/conflicting relationships among 3H dimensions, (ii) the consistent superiority of model merging over data mixture approaches in balancing alignment trade-offs, and (iii) the critical role of parameter-level conflict resolution through redundant component pruning and outlier mitigation. We propose R-TSVM, a Reweighting-enhanced Task Singular Vector Merging method that incorporates outlier-aware parameter weighting and sparsity-adaptive rank selection strategies adapted to the heavy-tailed parameter
• Score: 35.53877806259048
• License:
• Abstract: Achieving balanced alignment of large language models (LLMs) in terms of Helpfulness, Honesty, and Harmlessness (3H optimization) constitutes a cornerstone of responsible AI, with existing methods like data mixture strategies facing limitations including reliance on expert knowledge and conflicting optimization signals. While model merging offers a promising alternative by integrating specialized models, its potential for 3H optimization remains underexplored. This paper establishes the first comprehensive benchmark for model merging in 3H-aligned LLMs, systematically evaluating 15 methods (12 training-free merging and 3 data mixture techniques) across 10 datasets associated with 5 annotation dimensions, 2 LLM families, and 2 training paradigms. Our analysis reveals three pivotal insights: (i) previously overlooked collaborative/conflicting relationships among 3H dimensions, (ii) the consistent superiority of model merging over data mixture approaches in balancing alignment trade-offs, and (iii) the critical role of parameter-level conflict resolution through redundant component pruning and outlier mitigation. Building on these findings, we propose R-TSVM, a Reweighting-enhanced Task Singular Vector Merging method that incorporates outlier-aware parameter weighting and sparsity-adaptive rank selection strategies adapted to the heavy-tailed parameter distribution and sparsity for LLMs, further improving LLM alignment across multiple evaluations. We release our trained models for further exploration.

Related papers

• Reward-Guided Speculative Decoding for Efficient LLM Reasoning [80.55186052123196]
  We introduce Reward-Guided Speculative Decoding (RSD), a novel framework aimed at improving the efficiency of inference in large language models (LLMs) RSD incorporates a controlled bias to prioritize high-reward outputs, in contrast to existing speculative decoding methods that enforce strict unbiasedness. RSD delivers significant efficiency gains against decoding with the target model only, while achieving significant better accuracy than parallel decoding method on average.
  arXiv  Detail & Related papers  (2025-01-31T17:19:57Z)
• The Dual-use Dilemma in LLMs: Do Empowering Ethical Capacities Make a Degraded Utility? [54.18519360412294]
  Large Language Models (LLMs) must balance between rejecting harmful requests for safety and accommodating legitimate ones for utility. This paper presents a Direct Preference Optimization (DPO) based alignment framework that achieves better overall performance. Our resulting model, LibraChem, outperforms leading LLMs including Claude-3, GPT-4o, and LLaMA-3 by margins of 13.44%, 7.16%, and 7.10% respectively.
  arXiv  Detail & Related papers  (2025-01-20T06:35:01Z)
• Enhancing Few-Shot Learning with Integrated Data and GAN Model Approaches [35.431340001608476]
  This paper presents an innovative approach to enhancing few-shot learning by integrating data augmentation with model fine-tuning. It aims to tackle the challenges posed by small-sample data in fields such as drug discovery, target recognition, and malicious traffic detection. Results confirm that the MhERGAN algorithm developed in this research is highly effective for few-shot learning.
  arXiv  Detail & Related papers  (2024-11-25T16:51:11Z)
• Model-GLUE: Democratized LLM Scaling for A Large Model Zoo in the Wild [84.57103623507082]
  This paper introduces Model-GLUE, a holistic Large Language Models scaling guideline. We benchmark existing scaling techniques, especially selective merging, and variants of mixture. We then formulate an optimal strategy for the selection and aggregation of a heterogeneous model zoo. Our methodology involves the clustering of mergeable models and optimal merging strategy selection, and the integration of clusters.
  arXiv  Detail & Related papers  (2024-10-07T15:55:55Z)
• Making Large Language Models Better Planners with Reasoning-Decision Alignment [70.5381163219608]
  We motivate an end-to-end decision-making model based on multimodality-augmented LLM. We propose a reasoning-decision alignment constraint between the paired CoTs and planning results. We dub our proposed large language planners with reasoning-decision alignment as RDA-Driver.
  arXiv  Detail & Related papers  (2024-08-25T16:43:47Z)
• Progressively Label Enhancement for Large Language Model Alignment [42.01694160556464]
  Large Language Models (LLM) alignment aims to prevent models from producing content that misaligns with human expectations. We propose PLE, a framework that dynamically adjusts the model's training process based on the evolving quality of the generated data.
  arXiv  Detail & Related papers  (2024-08-05T16:21:17Z)
• Model Merging and Safety Alignment: One Bad Model Spoils the Bunch [70.614652904151]
  Merging Large Language Models (LLMs) is a cost-effective technique for combining multiple expert LLMs into a single versatile model. Current approaches often overlook the importance of safety alignment during merging, leading to highly misaligned models. We evaluate several popular model merging techniques, demonstrating that existing methods do not only transfer domain expertise but also propagate misalignment.
  arXiv  Detail & Related papers  (2024-06-20T17:59:58Z)
• Dataless Knowledge Fusion by Merging Weights of Language Models [51.8162883997512]
  Fine-tuning pre-trained language models has become the prevalent paradigm for building downstream NLP models. This creates a barrier to fusing knowledge across individual models to yield a better single model. We propose a dataless knowledge fusion method that merges models in their parameter space.
  arXiv  Detail & Related papers  (2022-12-19T20:46:43Z)
• RoCourseNet: Distributionally Robust Training of a Prediction Aware Recourse Model [29.057300578765663]
  RoCourseNet is a training framework that jointly optimize predictions and recourses that are robust to future data shifts. We show that RoCourseNet consistently achieves more than 96% robust validity and outperforms state-of-the-art baselines by at least 10% in generating robust explanations.
  arXiv  Detail & Related papers  (2022-06-01T18:18:18Z)
• Learning Distributionally Robust Models at Scale via Composite Optimization [45.47760229170775]
  We show how different variants of DRO are simply instances of a finite-sum composite optimization for which we provide scalable methods. We also provide empirical results that demonstrate the effectiveness of our proposed algorithm with respect to the prior art in order to learn robust models from very large datasets.
  arXiv  Detail & Related papers  (2022-03-17T20:47:42Z)

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.