SCOUT matches full-attention transformer scores on long-context language modeling and reasoning benchmarks at 400M and 1.3B scales while attending only to strided checkpoint tokens.
https://www.anthropic.com/index/ introducing-claude
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
-
SCOUT: Toward Sub-Quadratic Attention via Segment Compression for Optimized Utility in Transformers
SCOUT matches full-attention transformer scores on long-context language modeling and reasoning benchmarks at 400M and 1.3B scales while attending only to strided checkpoint tokens.