Diagonal Batching regroups layer-segment computations into parallel diagonals, preserving exact recurrence and speeding up long-context recurrent memory inference without retraining.
Transformer-XL: Attentive language models beyond a fixed-length context
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.LG 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Diagonal Batching Unlocks Parallelism in Recurrent Memory Transformers for Long Contexts
Diagonal Batching regroups layer-segment computations into parallel diagonals, preserving exact recurrence and speeding up long-context recurrent memory inference without retraining.