PulseAugur
中
实时 14:29:50
English(EN) 769 tok/s on a Single RTX 4090: Spark-X2.6-1.7B DSpark

新型LLM Spark-X2.6-1.7B在RTX 4090上实现769 tokens/sec

作者 PulseAugur 编辑部 · [1 个来源] ·

一款名为Spark-X2.6-1.7B的新语言模型已发布，并配备了名为DSpark的配套技术，在单块NVIDIA RTX 4090 GPU上实现了每秒769个token的吞吐量。该性能是在GSM8K基准测试上使用llama.cpp的Q4_K_M量化实现的。DSpark技术通过并行验证多个草稿token，实现了高达2.43倍的BF16精度加速，提高了效率。该模型可在Hugging Face和ModelScope上根据Apache 2.0许可证下载。 AI

影响 此次发布展示了小型模型在推理速度方面取得的显著进步，有可能实现更高效的本地部署和使用。

排序理由 发布一款具有性能基准和相关技术的新语言模型。[lever_c_demoted from research: ic=1 ai=1.0]

在 dev.to — LLM tag 阅读 →

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

新型LLM Spark-X2.6-1.7B在RTX 4090上实现769 tokens/sec

本文如何被排名

Signal score
1 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
发布一款具有性能基准和相关技术的新语言模型。[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
model release, 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
2 days old
Coverage has settled into its steady-state source set.

完整方法见我们的编辑标准。

报道来源 [1]

  1. dev.to — LLM tag TIER_1 English(EN) · SparkLLM ·

    769 tok/s on a Single RTX 4090: Spark-X2.6-1.7B DSpark

    <p><strong>769 tokens per second on a single NVIDIA RTX 4090.</strong> Spark-X2.6-1.7B with DSpark reaches that peak in the published GSM8K benchmark using a Q4_K_M target with llama.cpp, up from 424 tok/s without DSpark.</p> <p>For BF16, the highest reported speedup is <strong>2…

相关实体

相关话题