pith. sign in

13 Online V ector Quantized Attention

Next, we show that, under this initialization scheme, our assumption, Σn =I 1 β , the first batch EM step on a GMM is equivalent to a batch k-means update, we explain how ou

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it
browse 1 citing papers

fields

cs.LG 1

years

2026 1

verdicts

UNVERDICTED 1

representative citing papers

Online Vector Quantized Attention

cs.LG · 2026-02-03 · unverdicted · novelty 6.0

OVQ-attention delivers linear-time constant-memory sequence mixing via sparse Gaussian-mixture-based memory updates, matching self-attention performance on tasks up to 64k length while using far less memory.

citing papers explorer

Showing 1 of 1 citing paper.

  • Online Vector Quantized Attention cs.LG · 2026-02-03 · unverdicted · none · ref 6

    OVQ-attention delivers linear-time constant-memory sequence mixing via sparse Gaussian-mixture-based memory updates, matching self-attention performance on tasks up to 64k length while using far less memory.