# Mediator: Memory-efficient LLM Merging with Less Parameter Conflicts and Uncertainty Based Routing

arXiv: https://arxiv.org/abs/2502.04411 · HTML: https://mlsanity.com/paper/2502.04411/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 6 Feb 2025.

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

- [Can Compressed LLMs Truly Act? An Empirical Evaluation of Agentic Capabilities in LLM Compression](https://mlsanity.com/paper/2505.19433/) (2025)
- [Local Mixtures of Experts: Essentially Free Test-Time Training via Model Merging](https://mlsanity.com/paper/2505.14136/) (2025)
- [One-shot Federated Learning Methods: A Practical Guide](https://mlsanity.com/paper/2502.09104/) (2025)
- [Layer-Aware Task Arithmetic: Disentangling Task-Specific and Instruction-Following Knowledge](https://mlsanity.com/paper/2502.20186/) (2025)
- [PC-MoE: Memory-Efficient and Privacy-Preserving Collaborative Training for Mixture-of-Experts LLMs](https://mlsanity.com/paper/2506.02965/) (2025)
- [Making Bias Non-Predictive: Training Robust LLM Reasoning via Reinforcement Learning](https://mlsanity.com/paper/2602.01528/) (2026)
- [Can LoRA Fusion Support Cross-Domain Tasks in Cloud-Edge Collaboration?](https://mlsanity.com/paper/2605.23913/) (2026)

## Abstract (arXiv)

Model merging aggregates Large Language Models (LLMs) finetuned on different tasks into a stronger one. However, parameter conflicts between models leads to performance degradation in averaging. While model routing addresses this issue by selecting individual models during inference, it imposes excessive storage and compute costs, and fails to leverage the common knowledge from different models. In this work, we observe that different layers exhibit varying levels of parameter conflicts. Building on this insight, we average layers with minimal parameter conflicts and use a novel task-level expert routing for layers with significant conflicts. To further reduce storage costs, inspired by task arithmetic sparsity, we decouple multiple fine-tuned experts into a dense expert and several sparse experts. Considering the out-of-distribution samples, we select and merge appropriate experts based on the task uncertainty of the input data. We conduct extensive experiments on both LLaMA and Qwen with varying parameter scales, and evaluate on real-world reasoning tasks. Results demonstrate that our method consistently achieves significant performance improvements while requiring less system cost.