Rewiring Experts on the Fly:Continuous Rerouting for Better Online Adaptation in Mixture-of-Expert models

• URL: http://arxiv.org/abs/2510.14853v1
• Date: Thu, 16 Oct 2025 16:24:36 GMT
• Title: Rewiring Experts on the Fly:Continuous Rerouting for Better Online Adaptation in Mixture-of-Expert models
• Authors: Guinan Su, Yanwu Yang, Li Shen, Lu Yin, Shiwei Liu, Jonas Geiping,
• Abstract summary: Mixture-of-Experts (MoE) models achieve efficient scaling through sparse expert activation, but often suffer from suboptimal routing decisions due to distribution shifts in deployment.<n>We propose textita data-free, online test-time framework that continuously adapts MoE routing decisions during text generation without external supervision or data.
• Score: 52.502867924372275
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Mixture-of-Experts (MoE) models achieve efficient scaling through sparse expert activation, but often suffer from suboptimal routing decisions due to distribution shifts in deployment. While existing test-time adaptation methods could potentially address these issues, they primarily focus on dense models and require access to external data, limiting their practical applicability to MoE architectures. However, we find that, instead of relying on reference data, we can optimize MoE expert selection on-the-fly based only on input context. As such, we propose \textit{a data-free, online test-time framework} that continuously adapts MoE routing decisions during text generation without external supervision or data. Our method cycles between two phases: During the prefill stage, and later in regular intervals, we optimize the routing decisions of the model using self-supervision based on the already generated sequence. Then, we generate text as normal, maintaining the modified router until the next adaption. We implement this through lightweight additive vectors that only update router logits in selected layers, maintaining computational efficiency while preventing over-adaptation. The experimental results show consistent performance gains on challenging reasoning tasks while maintaining robustness to context shifts. For example, our method achieves a 5.5\% improvement on HumanEval with OLMoE. Furthermore, owing to its plug-and-play property, our method naturally complements existing test-time scaling techniques, e.g., achieving 6\% average gains when incorporated with self-consistency on DeepSeek-V2-Lite.

Related papers

• Audio-Visual Continual Test-Time Adaptation without Forgetting [28.411642462831647]
  continual test-time adaptation involves continually adapting a source audio-visual model at test-time.<n>We propose a method, $texttAV-CTTA$, that improves test-time performance of the models without access to any source data.
  arXiv  Detail & Related papers  (2026-02-20T04:55:01Z)
• AceGRPO: Adaptive Curriculum Enhanced Group Relative Policy Optimization for Autonomous Machine Learning Engineering [52.67783579040657]
  AceGRPO is a machine learning system that prioritizes tasks at the agent's learning frontier to maximize learning efficiency.<n>Our trained Ace-30B model achieves a 100% valid submission rate on MLE-Bench-Lite, approaches the performance of proprietary frontier models, and outperforms larger open-source baselines.
  arXiv  Detail & Related papers  (2026-02-08T10:55:03Z)
• MPRU: Modular Projection-Redistribution Unlearning as Output Filter for Classification Pipelines [23.370444162993707]
  We propose an emphinductive approach to machine unlearning (MU)<n>Unlearning can be done by reversing the last training sequence. This is implemented by appending a projection-redistribution layer in the end of the model.<n>Experiment results show consistently similar output to a fully retrained model with a high computational cost reduction.
  arXiv  Detail & Related papers  (2025-10-30T08:09:37Z)
• ETTA: Efficient Test-Time Adaptation for Vision-Language Models through Dynamic Embedding Updates [5.84817561920117]
  Test-Time Adaptation adapts vision-language models to unlabeled test data in new domains.<n>Current cache-based TTA models store only a limited set of high-confidence samples.<n>We propose a Recursive Updating module that integrates all incoming test samples.
  arXiv  Detail & Related papers  (2025-08-07T23:11:33Z)
• Aligning Frozen LLMs by Reinforcement Learning: An Iterative Reweight-then-Optimize Approach [65.6966065843227]
  Iterative Reweight-then-IRO is a framework that performs RL-style alignment of a frozen base model without touching its parameters.<n>At test time, the value functions are used to guide the base model generation via a search-based optimization process.<n> Notably, users can apply IRO to align a model on their own dataset, similar to OpenAI's reinforcement fine-tuning (RFT)
  arXiv  Detail & Related papers  (2025-06-21T21:49:02Z)
• MoTE: Mixture of Task-specific Experts for Pre-Trained ModelBased Class-incremental Learning [39.892628170627496]
  Class-incremental learning (CIL) requires deep learning models to continuously acquire new knowledge from streaming data.<n> prompt-based approaches suffer from prompt overwriting, while adapter-based methods face challenges such as dimensional misalignment between tasks.<n>We propose a mixture of task-specific experts (MoTE) framework that effectively mitigates the miscalibration caused by inconsistent output dimensions.
  arXiv  Detail & Related papers  (2025-05-21T03:06:10Z)
• Entropy-Based Adaptive Weighting for Self-Training [15.089334734753677]
  We propose Entropy-Based Adaptive Weighting for Self-Training (EAST)<n>EAST is an adaptive weighting strategy designed to prioritize uncertain data during self-training.<n>We evaluate our approach on GSM8K and MATH benchmarks.
  arXiv  Detail & Related papers  (2025-03-31T10:04:35Z)
• Self-Augmented Preference Optimization: Off-Policy Paradigms for Language Model Alignment [104.18002641195442]
  We introduce Self-Augmented Preference Optimization (SAPO), an effective and scalable training paradigm that does not require existing paired data. Building on the self-play concept, which autonomously generates negative responses, we further incorporate an off-policy learning pipeline to enhance data exploration and exploitation.
  arXiv  Detail & Related papers  (2024-05-31T14:21:04Z)
• Fast-Slow Test-Time Adaptation for Online Vision-and-Language Navigation [67.18144414660681]
  We propose a Fast-Slow Test-Time Adaptation (FSTTA) approach for online Vision-and-Language Navigation (VLN) Our method obtains impressive performance gains on four popular benchmarks.
  arXiv  Detail & Related papers  (2023-11-22T07:47:39Z)
• AR-TTA: A Simple Method for Real-World Continual Test-Time Adaptation [1.4530711901349282]
  We propose to validate test-time adaptation methods using datasets for autonomous driving, namely CLAD-C and SHIFT. We observe that current test-time adaptation methods struggle to effectively handle varying degrees of domain shift. We enhance the well-established self-training framework by incorporating a small memory buffer to increase model stability.
  arXiv  Detail & Related papers  (2023-09-18T19:34:23Z)
• StableMoE: Stable Routing Strategy for Mixture of Experts [109.0602120199226]
  Mixture-of-Experts (MoE) technique can scale up the model size of Transformers with an affordable computational overhead. We propose StableMoE with two training stages to address the routing fluctuation problem. Results show that StableMoE outperforms existing MoE methods in terms of both convergence speed and performance.
  arXiv  Detail & Related papers  (2022-04-18T16:48:19Z)

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.