PulseAugur
中
实时 06:56:49
English(EN) The State Is Not the Bottleneck: A Paged-State Gated-DeltaNet Tile on a Tri-Mode Multicast Mesh, Bit-Exact from RTL to Silicon

FPGA硬件瓦片高效管理大语言模型状态

研究人员为FPGA开发了一种硬件瓦片,可以高效处理线性注意力语言模型的递归状态。该瓦片专为Agilex 5 FPGA上的3.69亿参数Gated DeltaNet模型设计,通过将其分页到外部内存来管理超出设备片上内存的递归状态。研究发现,此分页过程仅占每个token的解码时间的13%,表明其在边缘部署中的实用性。在该FPGA上该模型的性能瓶颈被确定为模型权重的传输,而非状态分页。 AI

影响 该研究展示了一种在内存有限的边缘设备上部署大语言模型的实用方法。

排序理由 这是一篇详细介绍语言模型新颖硬件实现的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.LG 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →

FPGA硬件瓦片高效管理大语言模型状态

本文如何被排名

Signal score
37 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
这是一篇详细介绍语言模型新颖硬件实现的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
paper, infra
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

  1. arXiv cs.LG TIER_1 English(EN) · Cong Xu, Ravi Sankar ·

    状态并非瓶颈:三模组播网格上的分页状态门控 DeltaNet 瓦片,从 RTL 到硅的位精确

    arXiv:2611.18422v1 Announce Type: cross Abstract: Linear-attention language models replace a growing key-value cache with a recurrent state of fixed size. On an edge FPGA, this state can exceed the on-chip memory and must be paged to external memory. This paper asks whether such …