PulseAugur
中
实时 11:08:42
English(EN) Reasoning Models Don't Just Think Longer, They Move Differently

经过推理训练的大型语言模型展现出超越生成长度的独特内部轨迹

研究人员开发了一种分析经过推理训练的语言模型的内部轨迹的方法,区分了仅仅是花费更多步骤和遵循不同的计算路径。通过调整生成长度,他们发现模型难度与修正后的轨迹几何形状相关,尤其是在编码任务中,与标准的指令微调模型相比,更难的问题在推理模型中显示出更直接的路径。在数学和布尔可满足性问题中也观察到了这种区别,尽管不太明显,这表明推理训练确实可以改变模型的内部处理方式,而不仅仅是长度。 AI

影响 提供了一种分析大型语言模型推理的新方法,有望带来更好的模型可解释性和有针对性的训练改进。

排序理由 该集群包含一篇学术论文,详细介绍了一种分析大型语言模型行为的新研究方法。

在 arXiv cs.CL 阅读 →

AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →

经过推理训练的大型语言模型展现出超越生成长度的独特内部轨迹

本文如何被排名

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Research
该集群包含一篇学术论文,详细介绍了一种分析大型语言模型行为的新研究方法。
Source corroboration
2 independent sources
Multiple independent publishers reporting the same story raises confidence that it's real and newsworthy.
Topics
paper, model release
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
139 days old
Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

完整方法见我们的编辑标准。

报道来源 [2]

  1. arXiv cs.CL TIER_1 English(EN) · Sanmi Koyejo ·

    推理模型不仅思考时间更长,它们的运作方式也不同

    Reasoning-trained language models often spend more tokens on harder problems, but longer chains of thought do not show whether a model is merely computing for more steps or following a different internal trajectory. We study this distinction through hidden-state trajectories duri…

  2. arXiv stat.ML TIER_1 English(EN) · Anders Gj{\o}lbye, Lars Kai Hansen, Sanmi Koyejo ·

    推理模型不仅思考时间更长,它们的运作方式也不同

    arXiv:2605.15454v1 Announce Type: cross Abstract: Reasoning-trained language models often spend more tokens on harder problems, but longer chains of thought do not show whether a model is merely computing for more steps or following a different internal trajectory. We study this …