Evals is a framework for evaluating OpenAI models and an open-source registry of benchmarks.
conda install conda-forge::evals