# LED-Merging: Mitigating Safety-Utility Conflicts in Model Merging with Location-Election-Disjoint

arXiv: https://arxiv.org/abs/2502.16770 · HTML: https://mlsanity.com/paper/2502.16770/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Builds on 51 of its 79 references that are in the map.
- First posted on arXiv 24 Feb 2025.

## Builds on (most cited first, 30 of 51)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Measuring Mathematical Problem Solving With the MATH Dataset](https://mlsanity.com/paper/2103.03874/) (2021)
- [Program Synthesis with Large Language Models](https://mlsanity.com/paper/2108.07732/) (2021)
- [Axiomatic Attribution for Deep Networks](https://mlsanity.com/paper/1703.01365/) (2017)
- [Learning Important Features Through Propagating Activation Differences](https://mlsanity.com/paper/1704.02685/) (2017)
- [HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal](https://mlsanity.com/paper/2402.04249/) (2024)
- [Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference time](https://mlsanity.com/paper/2203.05482/) (2022)
- [Inference-Time Intervention: Eliciting Truthful Answers from a Language Model](https://mlsanity.com/paper/2306.03341/) (2023)
- [WizardLM: Empowering large pre-trained language models to follow complex instructions](https://mlsanity.com/paper/2304.12244/) (2023)
- [SparseGPT: Massive Language Models Can Be Accurately Pruned in One-Shot](https://mlsanity.com/paper/2301.00774/) (2023)
- [Striving for Simplicity: The All Convolutional Net](https://mlsanity.com/paper/1412.6806/) (2014)
- [SimPO: Simple Preference Optimization with a Reference-Free Reward](https://mlsanity.com/paper/2405.14734/) (2024)
- [Are Sixteen Heads Really Better than One?](https://mlsanity.com/paper/1905.10650/) (2019)
- [SNIP: Single-shot Network Pruning based on Connection Sensitivity](https://mlsanity.com/paper/1810.02340/) (2018)
- [Discovering Latent Knowledge in Language Models Without Supervision](https://mlsanity.com/paper/2212.03827/) (2022)
- [Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunch](https://mlsanity.com/paper/2311.03099/) (2023)
- [WizardMath: Empowering Mathematical Reasoning for Large Language Models via Reinforced Evol-Instruct](https://mlsanity.com/paper/2308.09583/) (2023)
- [Merging Models with Fisher-Weighted Averaging](https://mlsanity.com/paper/2111.09832/) (2021)
- [The WMDP Benchmark: Measuring and Reducing Malicious Use With Unlearning](https://mlsanity.com/paper/2403.03218/) (2024)
- [The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructions](https://mlsanity.com/paper/2404.13208/) (2024)
- [Trustworthy LLMs: a Survey and Guideline for Evaluating Large Language Models' Alignment](https://mlsanity.com/paper/2308.05374/) (2023)
- [Dataless Knowledge Fusion by Merging Weights of Language Models](https://mlsanity.com/paper/2212.09849/) (2022)
- [LoraHub: Efficient Cross-Task Generalization via Dynamic LoRA Composition](https://mlsanity.com/paper/2307.13269/) (2023)
- [Label Words are Anchors: An Information Flow Perspective for Understanding In-Context Learning](https://mlsanity.com/paper/2305.14160/) (2023)
- [Assessing the Brittleness of Safety Alignment via Pruning and Low-Rank Modifications](https://mlsanity.com/paper/2402.05162/) (2024)

## Built on by (most cited first, 7 of 7)

- [Paper2Rebuttal: A Multi-Agent Framework for Transparent Author Response Assistance](https://mlsanity.com/paper/2601.14171/) (2026)
- [Human-Agent Collaborative Paper-to-Page Crafting](https://mlsanity.com/paper/2510.19600/) (2025)
- [Beyond Parameter Arithmetic: Sparse Complementary Fusion for Distribution-Aware Model Merging](https://mlsanity.com/paper/2602.11717/) (2026)
- [A Data-Efficient Path to Multilingual LLMs: Language Expansion via Post-training PARAMDelta Integration into Upcycled MoE](https://mlsanity.com/paper/2605.18083/) (2026)
- [Asymmetric Collapse in Model Merging: When Refusal Over- writes Recognition](https://mlsanity.com/paper/2607.27240/) (2026)
- [When Skills Meet Safety: Benchmarking and Characterizing the Adaptive Jailbreak Robustness of Skill-Merged LLMs](https://mlsanity.com/paper/2608.08542/) (2026)
- [DiffSafeMerge: Mitigating Backdoor Inheritance in Diffusion Model Merging](https://mlsanity.com/paper/2608.09445/) (2026)

## Abstract (arXiv)

Fine-tuning pre-trained Large Language Models (LLMs) for specialized tasks incurs substantial computational and data costs. While model merging offers a training-free solution to integrate multiple task-specific models, existing methods suffer from safety-utility conflicts where enhanced general capabilities degrade safety safeguards. We identify two root causes: neuron misidentification due to simplistic parameter magnitude-based selection, and cross-task neuron interference during merging. To address these challenges, we propose LED-Merging, a three-stage framework that Locates task-specific neurons via gradient-based attribution, dynamically Elects critical neurons through multi-model importance fusion, and Disjoints conflicting updates through parameter isolation. Extensive experiments on Llama-3-8B, Mistral-7B, and Llama2-13B demonstrate that LED-Merging effectively reduces harmful response rates, showing a 31.4% decrease on Llama-3-8B-Instruct on HarmBench, while simultaneously preserving 95% of utility performance, such as achieving 52.39% accuracy on GSM8K. LED-Merging resolves safety-utility conflicts and provides a lightweight, training-free paradigm for constructing reliable multi-task LLMs. Code is available at hrefhttps://github.com/MqLeet/LED-MergingGitHub.