CCA-Attention pools token groups into core tokens and adds a local window to replace self-attention, claiming long-context speedups with minimal fine-tuning.
Same as demonstrated in existing methods (Beltagy et al., 2020; Xiao et al., 2024b)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Core Context Aware Transformers for Long Context Language Modeling
CCA-Attention pools token groups into core tokens and adds a local window to replace self-attention, claiming long-context speedups with minimal fine-tuning.