The Hidden Attention of Mamba Models

arXiv:2403.01590v1 Announce Type: new
Abstract: The Mamba layer offers an efficient selective state space model (SSM) that is highly effective in modeling multiple domains including NLP, long-range sequences processing, and computer vision. Selective SSMs are viewed as dual models, in which one trains in parallel on the entire sequence via IO-aware parallel scan, and deploys in an autoregressive manner. We add a third view and show that such models can be viewed as attention-driven models. This new perspective enables us …

abstract arxiv attention computer computer vision cs.lg domains hidden layer mamba modeling multiple nlp processing space state trains type via vision