R2-T2: Re-Routing in Test-Time for Multimodal Mixture-of-Experts

TL;DR

This paper introduces R2-T2, a test-time re-routing method for multimodal mixture-of-experts models that locally optimizes routing weights to enhance performance on diverse tasks without retraining the base model.

Contribution

The paper presents a novel test-time re-routing approach (R2-T2) that improves multimodal MoE model performance by locally optimizing routing weights without additional training.

Findings

R2-T2 significantly boosts model performance on various benchmarks.

The method improves routing efficiency and task adaptability.

It achieves state-of-the-art results without retraining base models.

Abstract

In large multimodal models (LMMs), the perception of non-language modalities (e.g., visual representations) is usually not on par with the large language models (LLMs)' powerful reasoning capabilities, deterring LMMs' performance on challenging downstream tasks. This weakness has been recently mitigated by replacing the vision encoder with a mixture-of-experts (MoE), which provides rich, multi-granularity, and diverse representations required by diverse downstream tasks. The performance of multimodal MoE largely depends on its router, which reweights and mixes the representations of different experts for each input. However, we find that the end-to-end trained router does not always produce the optimal routing weights for every test sample. To bridge the gap, we propose a novel and efficient method "Re-Routing in Test-Time (R2-T2)" that locally optimizes the vector of routing weights in test-time by moving it toward those vectors of the correctly predicted samples in a neighborhood of the test sample. We propose three R2-T2 strategies with different optimization objectives and neighbor-search spaces. R2-T2 consistently and greatly improves state-of-the-art LMMs' performance on challenging benchmarks of diverse tasks, without training any base-model parameters.