English(EN) Objective vs. Search: Decomposing What Makes a Good Tokeniser

新研究解构语言模型的分词算法

作者 PulseAugur 编辑部 · [1 个来源] · 2026-09-17 04:00

研究人员引入了两种新的分词算法，BottomUpLL 和 TopDownComp，以解构语言模型分词器中优化目标和搜索程序的影响。通过创建一个 2x2 的设计空间，他们比较了自下而上和自上而下的方法与压缩和对数似然目标。他们的发现表明，搜索程序而不是目标是性能的主要驱动因素，自下而上的分词器通常能实现更低的每字节比特数。然而，在 BLiMP 任务上，设计选择与性能之间没有发现一致的关系。 AI

影响为更原则性地构建分词器提供了指导，有可能提高语言模型的效率和性能。

排序理由详细介绍新算法和实验结果的学术论文。[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 →

本文如何被排名

Signal score: 11 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 详细介绍新算法和实验结果的学术论文。[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, 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: Same-day; Cluster formed today. Ranking reflects the current source set at time of score.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.AI TIER_1 English(EN) · Ahmetcan Yavuz, Clara Meister, Tiago Pimentel · 2026-09-17 04:00

Objective vs. Search: Decomposing What Makes a Good Tokeniser

arXiv:2609.19145v1 Announce Type: cross Abstract: Two dominant tokenisation algorithms are used by modern language models: byte-pair encoding (BPE) and UnigramLM. These differ along two orthogonal axes: their optimisation objective (compression vs. log-likelihood) and their searc…

本文如何被排名

报道来源 [1]

Objective vs. Search: Decomposing What Makes a Good Tokeniser

相关实体

相关话题