Metraipolis › archive › dc_64ddb7ca6f8b

LampAttention: Look-Ahead Mixed-Precision FlashAttention for Dedicated Accelerators — arXiv:2609.39361

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.39361
License (as archived)
CC0-1.0
Archived
2026-10-07T13:46:07.422Z
SHA-256 of the archived text
ddc3b1b34ff2bc373969260fda1430039543a888ec2e76f247061618dcc6ee94
Stable URL of this page
https://metraipolis.com/documento/dc_64ddb7ca6f8b

How to cite

LampAttention: Look-Ahead Mixed-Precision FlashAttention for Dedicated Accelerators — arXiv:2609.39361 — arXiv (en), https://arxiv.org/abs/2609.39361 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_64ddb7ca6f8b (Metraipolis archive dc_64ddb7ca6f8b, sha256 ddc3b1b34ff2bc373969260fda1430039543a888ec2e76f247061618dcc6ee94)
BibTeX
@misc{dc64ddb7ca6f8b,
  title = {LampAttention: Look-Ahead Mixed-Precision FlashAttention for Dedicated Accelerators — arXiv:2609.39361},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.39361},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_64ddb7ca6f8b; sha256 ddc3b1b34ff2bc373969260fda1430039543a888ec2e76f247061618dcc6ee94},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.39361. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

While most attention logits can be computed in low precision without degrading numerical stability, current attention kernels fail to exploit this phenomenon. We introduce a novel hardware-algorithm co-design in the form of mixed-precision FlashAttention. Our method accumulates key-query products and evaluates their exponentials in 8-bit formats, then adaptively identifies sensitive sub-blocks and recomputes them in 16-bit formats. We propose the specifications for a dedicated accelerator capable of executing this pipeline efficiently. Simulated experiments with Qwen3 and Gemma 3 show that rerouting a selective minority of sub-blocks to high precision is sufficient to recover the baseline model performance.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG) ; Numerical Analysis (math.NA).
Pagina del lavoro: https://arxiv.org/abs/2609.39361 — 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 2026-10-07T13:18:10.370Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).