SCOREOther
Political Compass Jev Benchmark
True Become False@trubecomefalse
This project benchmarks Jev 1.13.0 by running 12 passes on the Political Compass website's questions and scoring, varying ordering and prompting language to ensure consistent results.
Benchmarked @typesafeai 's Jev 1.13.0 Ran 12 passes on politicalcompass.org questions and scoring. Varied ordering and prompting language to reduce chance of bias and got a really consistent cluster. x.com/trubecomefalse/status/21013938034960