NABLA builds content-aware block-sparse attention masks by pooling queries and keys and thresholding their softmax CDF, giving near-baseline video quality at up to 2.7x inference speedup and 1.46x pretraining speedup.
V aswaniet al., ‘‘Attention is all you need,’’Advances in neural infor- mation processing systems, vol
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
NABLA: Neighborhood Adaptive Block-Level Attention
NABLA builds content-aware block-sparse attention masks by pooling queries and keys and thresholding their softmax CDF, giving near-baseline video quality at up to 2.7x inference speedup and 1.46x pretraining speedup.