OmniMamba: Efficient and Unified Multimodal Understanding and Generation via State Space Models
This work addresses efficiency and data requirements for researchers and practitioners in multimodal AI, representing a significant but incremental improvement over existing methods.
The paper tackles the problem of high computational complexity and data inefficiency in unified multimodal understanding and generation models by introducing OmniMamba, a linear-architecture-based model that achieves competitive performance with JanusFlow and surpasses Show-o on benchmarks while using 1,000 times fewer training data (2M image-text pairs) and offering up to a 119.2 times speedup and 63% GPU memory reduction for long-sequence generation.
Recent advancements in unified multimodal understanding and visual generation (or multimodal generation) models have been hindered by their quadratic computational complexity and dependence on large-scale training data. We present OmniMamba, the first linear-architecture-based multimodal generation model that generates both text and images through a unified next-token prediction paradigm. The model fully leverages Mamba-2's high computational and memory efficiency, extending its capabilities from text generation to multimodal generation. To address the data inefficiency of existing unified models, we propose two key innovations: (1) decoupled vocabularies to guide modality-specific generation, and (2) task-specific LoRA for parameter-efficient adaptation. Furthermore, we introduce a decoupled two-stage training strategy to mitigate data imbalance between two tasks. Equipped with these techniques, OmniMamba achieves competitive performance with JanusFlow while surpassing Show-o across benchmarks, despite being trained on merely 2M image-text pairs, which is 1,000 times fewer than Show-o. Notably, OmniMamba stands out with outstanding inference efficiency, achieving up to a 119.2 times speedup and 63% GPU memory reduction for long-sequence generation compared to Transformer-based counterparts. Code and models are released at https://github.com/hustvl/OmniMamba