IndustrySelected

Thinking Machines 与 Goodfire 合作研究 looped transformers 可解释性

Why we picked it

Thinking Machines 和 Goodfire 一起研究 looped transformers,Goodfire 负责看层与循环之间怎么传信息、用什么工具能看清内部机制。做可解释性或关注非标准 Transformer 架构的研究者可以关注,等他们放论文或工具再说。

Thinking Machines 宣布与可解释性研究公司 Goodfire 建立研究合作,研究方向是 looped transformers(循环 Transformer)。Goodfire 的工作聚焦两个问题:模型各层如何把信息传递给后续循环,以及哪些可解释性工具能帮助理解模型的内部运作机制。

Body

这条推文本身没有给出模型参数量、基准成绩或时间表等具体数据,属于研究合作的初步披露。looped transformers 是一种让模型重复经过同一组层多次的结构,与一次前向传播的标准 Transformer 不同,研究其信息传递方式需要借助机制可解释性方法。对关注可解释性与新架构的研究者来说,后续应留意两家公司是否会发布论文或开源工具。

Original · Thinking Machines Lab

We're grateful for our research partner working with us to study looped transformers.

Goodfire explored how layers pass information to future loops, and which interpretability tools allow us to understand the model's inner workings.

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Industry" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include Thinking Machines, Goodfire, looped transformers, interpretability.

Source attribution: curated from Thinking Machines Lab (@thinkymachines).