跳到正文
elvis· @omarsar0 · X·· 3 小时前AI 评分38
AI 导读

推荐阅读。我认为这是 OpenAI 的一项良好举措。如何衡量前沿 AI 以及对其形成共同、共享的理解至关重要。这使得评估进展、安全性等变得更加容易。 这不如完整的技术报告那么出色(令人遗憾行业在加速过程中迅速远离了这些报告),但它是迈向更多透明度的开端。随着我们向 RSI 迈进,这一点更为重要,因为测量将占据首要地位。

正文 · 原文

Recommended reading. I think it's a good initiative from OpenAI. How we measure frontier AI and a common, shared understanding of it are crucial. It makes it easier to assess progress, safety, etc.

It's not as great as full technical reports (sad how fast the industry moved away from those during the acceleration), but it's a start towards more transparency. This is even more important as we move toward RSI, where measurement takes top priority.

引用Jakub Pachocki@merettm
We decided to provide more transparency into pretraining progress at OpenAI for the sake of moving toward shared safety bars. In “Perplexity and Intelligence” I argue for common limits based on perplexity and other measurements, with progress beyond them gated on our ability to understand AI behavior. https://openai.com/index/perplexity-and-intelligence
在 X 查看被引用的帖子

来源:elvis · x.com