新闻
Perplexity and Intelligence
OpenAI · 发布于 · 阅读约3分钟 · Hacker News 100 分
30秒读懂
- 发生了什么
- OpenAI proposes perplexity, a pretraining metric measuring prediction accuracy, as a shared industry standard for tracking AI progress and establishing safety pacing limits.
- 为何重要
- Matters for AI researchers and safety teams evaluating frontier models and coordinating development pace across competing labs.
- 注意
- Perplexity doesn't directly measure real-world capabilities, varies across datasets and training data, and doesn't account for post-training improvements like reasoning.
- eval
这条新闻背后的模式
每个模式都讲清楚技术如何运作、何时值得投入,以及在哪里会失效。
The Agent Architect
每周一个模式、一个权衡、一个生产事故案例。为构建智能体系统的人准备的每周简报。
每周一封邮件,一键退订。您的地址仅用于发送简报。