IAAIFlashLoop
FlashLoop: Fast and Memory-Efficient Looped Transformers via Lazy Updates
A training-free inference framework that removes cross-loop redundancy: later loops update only the few tokens whose states still change, attend to a sparse stable set of key columns, and store KV residuals between adjacent loops in low bit-width.
Inside the method
Loop r: sparse token updates + sparse attention + quantized KV residual vs loop r−1
Simplified conceptual schematic. Consult the paper for the complete architecture.
- Recurrence family
- Full-stack recurrence
- Depth control
- Fixed loops of the base model
- KV / state strategy
- Low-bit KV residuals between adjacent loops
Reading note
Across several looped models it reports lossless accuracy with up to 1.64× end-to-end speedup and up to 6× KV-cache memory reduction.
Sources checked 2026-10-28. This catalog does not imply independent reproduction.
Related works
Cite this work
@misc{flashloop2026,
title = {FlashLoop: Fast and Memory-Efficient Looped Transformers via Lazy Updates},
author = {Wanqi Yang and Shiwei Liu},
year = {2026},
eprint = {2609.29812},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.29812}
}