Sparse Attention with Persistent State Machines – High‑Sparsity LLM Acceleratorzenodo.org3 points·yusuke_esaka··0 commentsOpen articleSaveView on HN