Mixture of States: Routing Token-Level Dynamics for Multimodal Generation

• URL: http://arxiv.org/abs/2511.12207v1
• Date: Sat, 15 Nov 2025 13:24:57 GMT
• Title: Mixture of States: Routing Token-Level Dynamics for Multimodal Generation
• Authors: Haozhe Liu, Ding Liu, Mingchen Zhuge, Zijian Zhou, Tian Xie, Sen He, Yukang Yang, Shuming Liu, Yuren Cong, Jiadong Guo, Hongyu Xu, Ke Xu, Kam-Woh Ng, Juan C. Pérez, Juan-Manuel~Pérez-Rúa, Tao Xiang, Wei Liu, Shikun Liu, Jürgen Schmidhuber,
• Abstract summary: We introduce MoS (Mixture of States), a novel fusion paradigm for multimodal diffusion models.<n>MoS is a learnable, token-wise router that creates denoising timestep- and input-dependent interactions.<n>We validate our design with text-to-image generation and editing, which achieve state-of-the-art results.
• Score: 60.15447534872979
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We introduce MoS (Mixture of States), a novel fusion paradigm for multimodal diffusion models that merges modalities using flexible, state-based interactions. The core of MoS is a learnable, token-wise router that creates denoising timestep- and input-dependent interactions between modalities' hidden states, precisely aligning token-level features with the diffusion trajectory. This router sparsely selects the top-$k$ hidden states and is trained with an $ε$-greedy strategy, efficiently selecting contextual features with minimal learnable parameters and negligible computational overhead. We validate our design with text-to-image generation (MoS-Image) and editing (MoS-Editing), which achieve state-of-the-art results. With only 3B to 5B parameters, our models match or surpass counterparts up to $4\times$ larger. These findings establish MoS as a flexible and compute-efficient paradigm for scaling multimodal diffusion models.

Related papers

• DM$^3$T: Harmonizing Modalities via Diffusion for Multi-Object Tracking [10.270441242480482]
  This paper proposes DM$3$T, a novel framework that reformulates multimodal fusion as an iterative feature alignment process.<n>Our approach performs iterative cross-modal harmonization through a proposed Cross-Modal Diffusion Fusion (C-MDF) module.<n>To further improve tracking robustness, we design a Hierarchical Tracker that adaptively handles confidence estimation.
  arXiv  Detail & Related papers  (2025-11-28T06:02:58Z)
• MoIIE: Mixture of Intra- and Inter-Modality Experts for Large Vision Language Models [52.876185634349575]
  We propose to incorporate Mixture of Intra- and Inter-Modality Experts (MoIIE) to Large Vision-Language Models (LVLMs)<n>For each token, expert routing is guided by its modality, directing tokens to their respective intra-modality experts as well as a shared pool of inter-modality experts.<n>Our MoIIE models with 5.5B and 11.3B activated parameters match or even surpass the performance of existing advanced open-source MoE-LLMs based multi-modal models.
  arXiv  Detail & Related papers  (2025-08-13T13:00:05Z)
• SMAR: Soft Modality-Aware Routing Strategy for MoE-based Multimodal Large Language Models Preserving Language Capabilities [13.808550562570275]
  Existing methods to build multimodal MoE models either incur high training costs or suffer from degraded language capabilities when adapting pretrained models.<n>We propose Soft ModalityAware Routing (SMAR), a novel regularization technique that uses Kullback Leibler divergence to control routing probability distributions.<n> Experiments on visual instruction tuning show that SMAR preserves language ability at 86.6% retention with only 2.5% pure text, outperforming baselines while maintaining strong multimodal performance.
  arXiv  Detail & Related papers  (2025-06-06T12:47:29Z)
• OptMerge: Unifying Multimodal LLM Capabilities and Modalities via Model Merging [124.91183814854126]
  Model merging seeks to combine multiple expert models into a single model.<n>We introduce a benchmark for model merging research that clearly divides the tasks for MLLM training and evaluation.<n>We find that model merging offers a promising way for building improved MLLMs without requiring training data.
  arXiv  Detail & Related papers  (2025-05-26T12:23:14Z)
• DiffMoE: Dynamic Token Selection for Scalable Diffusion Transformers [86.5541501589166]
  DiffMoE is a batch-level global token pool that enables experts to access global token distributions during training.<n>It achieves state-of-the-art performance among diffusion models on ImageNet benchmark.<n>The effectiveness of our approach extends beyond class-conditional generation to more challenging tasks such as text-to-image generation.
  arXiv  Detail & Related papers  (2025-03-18T17:57:07Z)
• MIFNet: Learning Modality-Invariant Features for Generalizable Multimodal Image Matching [54.740256498985026]
  Keypoint detection and description methods often struggle with multimodal data.<n>We propose a modality-invariant feature learning network (MIFNet) to compute modality-invariant features for keypoint descriptions in multimodal image matching.
  arXiv  Detail & Related papers  (2025-01-20T06:56:30Z)
• Layerwise Recurrent Router for Mixture-of-Experts [42.36093735411238]
  Mixture-of-Experts (MoE) architecture stands out for its ability to scale model size without significantly increasing training costs.<n>Current MoE models often display parameter inefficiency.<n>We introduce the Layerwise Recurrent Router for Mixture-of-Experts (RMoE)
  arXiv  Detail & Related papers  (2024-08-13T10:25:13Z)
• Lightweight In-Context Tuning for Multimodal Unified Models [57.10831399642176]
  MultiModal In-conteXt Tuning (M$2$IXT) is a lightweight module to enhance the ICL capabilities of multimodal unified models. When tuned on as little as 50K multimodal data, M$2$IXT can boost the few-shot ICL performance significantly.
  arXiv  Detail & Related papers  (2023-10-08T10:47:24Z)

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.