English(EN) Instruction Capacity: Behavior and Degradation of Language Models under Controlled Load

新工具衡量语言模型指令容量

作者 PulseAugur 编辑部 · [1 个来源] · 2026-10-26 00:00

研究人员开发了一种新颖的统计工具来衡量语言模型的指令容量，量化其在受控负载下的性能。该工具使用原子替换规则来评估模型在负载增加时遵循指令的准确性。对GPT-4o mini的初步测量表明，其指令容量在95%的准确率下约为37-38条规则，并且随着负载的增加，响应质量会下降。 AI

影响提供了一种新的定量方法来评估和比较不同语言模型的指令遵循能力。

排序理由该集群包含一篇详细介绍语言模型评估新方法的论文。

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: 0 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该集群包含一篇详细介绍语言模型评估新方法的论文。
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: 4 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。

Hugging Face Daily Papers TIER_1 English(EN) · 2026-10-26 00:00

指令容量：语言模型在受控负载下的行为与退化

Researchers developed a statistical instrument using atomic replacement rules to measure language model instruction capacity with quantified uncertainty, finding GPT-4o mini holds about 37-38 rules at 95% accuracy.