SpotAttention is a plug-in selector trained by KL distillation to route sparse attention in long-context transformers, matching dense accuracy up to 128K tokens with 3.9x decode speedup.
arXiv:2412.16187 , year =
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
SpotAttention: Plug-In Block-Sparse Routing for Pretrained Long-Context Transformers
SpotAttention is a plug-in selector trained by KL distillation to route sparse attention in long-context transformers, matching dense accuracy up to 128K tokens with 3.9x decode speedup.