PulseAugur
实时 21:07:46
English(EN) A Shared Subcircuit Lets LLMs Count Down Across Tasks

Llama-3.1-70B-Instruct 采用共享“倒计时子电路”来完成任务

研究人员在 Llama-3.1-70B-Instruct 中发现了一个“倒计时子电路”,使模型能够跟踪和管理特定任务的 token 数量。该子电路将当前位置与目标长度进行比较,估算完成剩余时间。研究发现,这种机制并非 Llama-3.1-70B-Instruct 所独有,因为在另一个前沿 LLM 中也曾观察到类似的模式,这表明不同模型之间存在共享组件。研究表明,通过逆向工程这些子电路,我们可以更好地理解特定行为如何在不同任务甚至不同模型之间泛化。 AI

影响 识别出 LLM 之间用于任务完成的共享机制,可能提高模型的解释性和泛化能力。

排序理由 论文详细介绍了在 LLM 中发现的特定子电路。[lever_c_demoted from research: ic=1 ai=1.0]

在 Hugging Face Daily Papers 阅读 →

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

Llama-3.1-70B-Instruct 采用共享“倒计时子电路”来完成任务

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
论文详细介绍了在 LLM 中发现的特定子电路。[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
53 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 [1]

  1. Hugging Face Daily Papers TIER_1 English(EN) ·

    共享子电路让大型语言模型跨任务倒计时

    Writing a sentence of exactly twelve words; ending a DNA sequence at the right codon; formatting an ASCII table. These are all tasks that language models can do that requires tracking how many tokens remain before a target. In this work, we identify in Llama-3.1-70B-Instruct a ge…