My benchmark for large language models

A benchmark of ~100 tests for language models, collected from actual questions I've asked of language models in the last year....