Metraipolis › archive › dc_db1f895348cf
FlashLoop: Fast and Memory-Efficient Looped Transformers via Lazy Updates — arXiv:2609.29812
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.29812
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-08T05:26:11.464Z
- SHA-256 of the archived text
7774b7009cad52c96236a5576515f22b7c880a009a3e4a52e6a4b0e2f4d4155d- Stable URL of this page
https://metraipolis.com/documento/dc_db1f895348cf
How to cite
FlashLoop: Fast and Memory-Efficient Looped Transformers via Lazy Updates — arXiv:2609.29812 — arXiv (en), https://arxiv.org/abs/2609.29812 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_db1f895348cf (Metraipolis archive dc_db1f895348cf, sha256 7774b7009cad52c96236a5576515f22b7c880a009a3e4a52e6a4b0e2f4d4155d)
BibTeX
@misc{dcdb1f895348cf,
title = {FlashLoop: Fast and Memory-Efficient Looped Transformers via Lazy Updates — arXiv:2609.29812},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.29812},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_db1f895348cf; sha256 7774b7009cad52c96236a5576515f22b7c880a009a3e4a52e6a4b0e2f4d4155d},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.29812. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Looped Transformers have attracted substantial attention as a parameter-efficient approach to increasing computational depth through repeated application of shared Transformer blocks. However, their practical advantages over conventional Transformers remain under debate: each additional loop incurs another Transformer pass and requires caching another set of KV states, causing inference FLOPs and KV-cache memory to grow continuously with loop depth. This overhead becomes particularly severe at large loop counts and long context, preventing the parameter efficiency of Looped Transformers from translating into practical inference efficiency. In this paper, we find that much of the additional computation and storage introduced by looping is redundant. As recurrence proceeds, state changes become increasingly concentrated on a small subset of tokens; attention-output differences are dominated by a sparse and stable subset of key columns; and KV residuals between adjacent loops become progressively more amenable to low-bit quantization. Building on these observations, we introduce FlashLoop, a training-free inference framework that reduces cross-loop redundancy through token-sparse updates, sparse attention, and KV-residual quantization. Across several Looped Transformers models, FlashLoop delivers lossless accuracy while achieving up to 1.64$\times$ end-to-end speedup and up to 6$\times$ KV-cache memory reduction, substantially improving the practicality of scaling Looped Transformers to greater computational depths and longer context. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-25 in cs.LG). Categorie arXiv: Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2609.29812 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-08T04:50:39.483Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-25?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).