Get the latest tech news

Bad benchmarks and evals: Senior SWE-Bench, napkin math, and winter tires


're going to look at three different kinds of benchmarks, one set of calculations for baseline numbers for performance "napkin math" estimates, one set of AI model evals, and one on car tires. To build my intuition for things, I like thinking about them before seeing the explanation, so these are presented with the benchmark information first and the explanation later in case you want to think about your answer before seeing my thoughts.

None

Get the Android app

Or read this on Hacker News

Read more on:

Photo of Napkin math

Napkin math

Photo of evals

evals

Photo of SWE-Bench

SWE-Bench

Related news:

News photo

AI Engineer Notebooks – free, framework-free RAG/agents/evals on Colab

News photo

Show HN: Product analytics (and evals) for agent sessions on your MCP

News photo

At Waymo, an AI project isn't ready until its evals are — not when the model performs well