PulseAugur
实时 08:21:07
English(EN) GGUF-Metadata Prediction of Single-Sequence llama.cpp Throughput Across Three Systems

新方法使用GGUF元数据预测llama.cpp吞吐量

研究人员开发了一种方法,利用GGUF元数据来预测llama.cpp(一个流行的运行大型语言模型的框架)的单序列吞吐量。该方法采用具有量化特定比例因子的屋顶线形预测器,并在参考模型上进行拟合。该研究在两个Apple M4 Max系统和一个NVIDIA RTX 5080上测试了53种配置下的318次测量,在某些留一主机测试中实现了低至11.6%的平均绝对百分比误差(MAPE)。研究结果表明,GGUF结构有助于跨不同系统的性能预测,尽管拟合的效率并非普遍适用。 AI

影响 提供了一种优化大型语言模型在本地硬件上部署和性能调优的方法。

排序理由 学术论文,详细介绍了一种预测模型吞吐量的新方法。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 阅读 →

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

新方法使用GGUF元数据预测llama.cpp吞吐量

本文如何被排名

Signal score
17 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
学术论文,详细介绍了一种预测模型吞吐量的新方法。[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.AI TIER_1 English(EN) · Xinyu Qiu, Chuhong Xu, Bo Su, Ziyao Chen, Ruiyang Xu, Shimeng Dai ·

    GGUF元数据预测llama.cpp单序列吞吐量跨三个系统

    arXiv:2609.14864v1 Announce Type: new Abstract: We predict single-sequence model throughput from GGUF metadata using roofline-shaped predictors with quantization-specific scale factors fitted on reference models. The scored cohort comprises 318 phase-depth measurements from 53 ho…