// TOPIC STREAM
State-space sequence models.
This article examines how the Structured State Space sequence model (S4) structures state space models to handle long dependencies and reduces them to computable Cauchy kernel operations. It distinguishes only the claims confirmed by the arXiv abstract and metadata.
Hungry Hungry Hippos decomposes the linguistic weaknesses of state space models into token recall and comparison, then explains how H3, FlashConv, and hybrid attention combinations address them.
This article explains why Mamba targets both long-sequence efficiency and language-modeling performance through selective SSMs that change state-space parameters based on the input and hardware-aware parallel algorithms.
Following Tri Dao and Albert Gu's Mamba-2 paper, this article explains the connection between state space models and attention, the hardware implications of the SSD algorithm, and the claims that remain unverified.
This article distinguishes fixed-size recurrent hidden states as memory devices versus information compression. It explains the role of state through S4, Mamba, RWKV, and Resona, while treating Qdrant and Statey separately as external-memory contexts.
This article reads the competition between state space models and Transformers as a question of division of labor rather than structural replacement. Based on the abstracts and publicly stated claims of Mamba, Griffin, Jamba, and Mamba-3, it organizes the trade-offs among long contexts, generation, and hardware efficiency.