AI evaluation, explained
How do we know an AI model is actually good?
GenAIBench is a place to learn how generative AI is measured: the benchmarks, the methods behind them, where they break, and what the research says.
Latest writing
All postsThe first posts are on their way.
What you'll find here
-
Tutorials
Step-by-step guides to running and building evaluations yourself.
0 posts
-
Articles
How benchmarks and metrics work, and where they break.
0 posts
-
Paper reviews
Careful readings of evaluation research, with an honest critique.
0 posts
-
Notes
Shorter observations on results, surprises and open questions.
0 posts