Every AI Benchmark I’ve Published Had a Bug That Would Have Flipped the Headline.
中文摘要
作者承认其发布的每个AI基准测试都存在足以改变结论的Bug。
English Summary
The author admits that every AI benchmark they published contained a bug that could have flipped the headline.
原文节选
Not a benchmark I read about. Every benchmark I built, measured, and put my name on. Continue reading on Medium »