VL-Mamba: Exploring State Space Models for Multimodal Learning (2403.13600v1)

Abstract: Multimodal LLMs (MLLMs) have attracted widespread interest and have rich applications. However, the inherent attention mechanism in its Transformer structure requires quadratic complexity and results in expensive computational overhead. Therefore, in this work, we propose VL-Mamba, a multimodal LLM based on state space models, which have been shown to have great potential for long-sequence modeling with fast inference and linear scaling in sequence length. Specifically, we first replace the transformer-based backbone LLM such as LLama or Vicuna with the pre-trained Mamba LLM. Then, we empirically explore how to effectively apply the 2D vision selective scan mechanism for multimodal learning and the combinations of different vision encoders and variants of pretrained Mamba LLMs. The extensive experiments on diverse multimodal benchmarks with competitive performance show the effectiveness of our proposed VL-Mamba and demonstrate the great potential of applying state space models for multimodal learning tasks.