Newest papers

Stratified Sparse Attention: Reducing the Sparse-to-Dense Gap at Extreme Sparsity

Sahil Joshi, Prithvi Dixit, Aryan Katiyar, Agniva Chowdhury, Joseph E. Gonzalez, Anshumali Shrivastava, Aditya Desai, Kumar Krishna Agrawal · 2026 · arXiv

DOI: 10.48550/arXiv.2610.28441 · open-access PDF

Portuguese

A inferência do modelo de idioma grande em contextos longos é cada vez mais bloqueada pelo custo de assistir a um cache de valor-chave (KV) crescente. A atenção reduz esse custo ao acessar apenas uma fração do cache, mas os métodos existentes enfrentam um compromisso entre a recuperação determinista, que captura tokens de alta pontuação, mas descartam a massa de atenção residual, e a amostragem, que melhora a cobertura ao custo de uma variância maior. Métodos híbridos combinam a recuperação de $k$ de topo com a amostragem uniforme, mas ainda exibem uma lacuna substancial entre a atenção densa e a escassez agressiva.

Machine translation (Meta M2M100); the English original follows.

Abstract

Large Language Model inference over long contexts is increasingly bottlenecked by the cost of attending to a growing Key-Value (KV) cache. Sparse attention reduces this cost by accessing only a fraction of the cache, but existing methods face a trade-off between deterministic retrieval, which captures high-scoring tokens but discards residual attention mass, and sampling, which improves coverage at the cost of higher variance. Hybrid methods combine top-$k$ retrieval with uniform sampling, yet still exhibit a substantial gap from dense attention at aggressive sparsity. We introduce \emph{Strat

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Is this paper real? This page was built from OpenAlex and arXiv records, not generated. Software can verify any reference in one call: GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.28441, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txt

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange