Evaluation

First Tragedy, then Parse: History Repeats Itself in the New Era of Large Language Models

Benchmarks as Microscopes: A Call for Model Metrology