# FedWeave: Rethinking the Unit of Specialization in Heterogeneous Federated MoE-LoRA

arXiv: https://arxiv.org/abs/2607.26618 · HTML: https://mlsanity.com/paper/2607.26618/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning.
- Builds on 14 of its 46 references that are in the map.
- First posted on arXiv 29 Jul 2026.

## Placed in

- LLM reasoning / [Test-time compute and efficient thinking](https://mlsanity.com/lr/test-time-compute-and-efficient-thinking/)
- Mixture-of-experts / [Federated mixture-of-experts](https://mlsanity.com/moe/federated-mixture-of-experts/)

## Builds on (most cited first, 14 of 14)

- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge](https://mlsanity.com/paper/1803.05457/) (2018)
- [Gemma 2: Improving Open Language Models at a Practical Size](https://mlsanity.com/paper/2408.00118/) (2024)
- [Towards Building the Federated GPT: Federated Instruction Tuning](https://mlsanity.com/paper/2305.05644/) (2023)
- [LLaVA-MoLE: Sparse Mixture of LoRA Experts for Mitigating Data Conflicts in Instruction Finetuning MLLMs](https://mlsanity.com/paper/2401.16160/) (2024)
- [Federated Mixture of Experts](https://mlsanity.com/paper/2107.06724/) (2021)
- [FedMoE: Personalized Federated Learning via Heterogeneous Mixture of Experts](https://mlsanity.com/paper/2408.11304/) (2024)
- [Specialized federated learning using a mixture of experts](https://mlsanity.com/paper/2010.02056/) (2020)
- [Personalized Federated Fine-tuning for Heterogeneous Data: An Automatic Rank Learning Approach via Two-Level LoRA](https://mlsanity.com/paper/2503.03920/) (2025)
- [Wireless Federated Multi-Task LLM Fine-Tuning via Sparse-and-Orthogonal LoRA](https://mlsanity.com/paper/2602.20492/) (2026)
- [FedTreeLoRA: Reconciling Statistical and Functional Heterogeneity in Federated LoRA Fine-Tuning](https://mlsanity.com/paper/2603.13282/) (2026)
- [Beyond Factor Aggregation: Gauge-Aware Low-Rank Server Representations for Federated LoRA](https://mlsanity.com/paper/2605.06733/) (2026)
- [SDFLoRA: Selective Decoupled Federated LoRA for Privacy-preserving Fine-tuning with Heterogeneous Clients](https://mlsanity.com/paper/2601.11219/) (2026)
- [Task-Centric Personalized Federated Fine-Tuning of Language Models](https://mlsanity.com/paper/2604.00050/) (2026)

## Abstract (arXiv)

Federated PEFT enables LLMs to collaboratively adapt to decentralized private data without sharing raw examples. However, task heterogeneity across clients can cause cross-task interference and gradient conflicts during aggregation. Federated MoE-LoRA addresses this challenge through specialized LoRA experts and conditional routing. Yet existing methods typically specialize at client granularity, implicitly assuming task-coherent clients. Our core insight is that experts need purity, namely pattern-coherent updates that preserve specialization, whereas routers need contrast, namely mixed-task observations that support expert comparison. We propose FedWeave, a framework that adopts asymmetric aggregation, separating expert aggregation from router optimization to meet these two requirements. FedWeave uses unsupervised prototype discovery to form local buckets and align them across clients, enabling prototype-level expert aggregation while retaining mixed-task client trajectories for router training. At inference, FedWeave performs sparse inference with one active expert while preserving nearly all soft-routing performance. Our theoretical analysis explains why asymmetric aggregation is advantageous: it controls expert convergence in stationarity through off-pattern contamination, identifies the consensus error induced by fragmented router trajectories, and bounds sparse-inference risk. On a heterogeneous multi-task benchmark with mainstream LLM backbones, FedWeave consistently outperforms strong baselines, while ablations verify the effectiveness of our design.