Tag: Multiple Comparisons

When we try many models, prompts, seeds, and metrics, false discoveries become easier than they look.

当我们测试很多模型、prompt、seed 和指标时，假阳性会比想象中更容易出现。