Papers & Research

Anthropic and other institutions reveal inverse scaling of test-time compute in large reasoning models

Original title:Inverse Scaling in Test-Time Compute

AI summary

Anthropic Fellows Program and other institutions built evaluation tasks and found that longer reasoning reduces the accuracy of large reasoning models, showing inverse scaling of test-time compute, and identified five failure modes.

Key Takeaways

  • Anthropic Fellows Program and other institutions built evaluation tasks and found that longer…
  • showing inverse scaling of test-time compute
  • and identified five failure modes.

Quick Facts

Published
Updated
Source
Anthropic:Alignment Science Blog(网页)
Category
Papers & Research
AI score
68
Read the original Read the full article at the original source
Why it matters

研究给出五类测试时计算反向扩展的失败模式,并附带安全含义,可帮助读者理解长推理并非总是增益。

Related entities

Source & copyright

This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.

Original article · AIHOT entry

How to Cite

When citing this article, keep the title, date, and permalink, and credit Orange News as the source.

Orange News: Anthropic and other institutions reveal inverse scaling of test-time compute in large reasoning models. July 23, 2025. https://new.nat6.net/en/news/pfau6cm0y08yi1y2alvpxdj79/

View all →