DistrAttention approximates self-attention by LSH-based grouping of embedding-dimension columns of Q and K, reducing compute along d while keeping all tokens in context.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
DistrAttention: An Efficient and Flexible Self-Attention Mechanism on Modern GPUs
DistrAttention approximates self-attention by LSH-based grouping of embedding-dimension columns of Q and K, reducing compute along d while keeping all tokens in context.