JevBench

JevBench is a benchmark for the abilities the benchmark was built to measure. The official description says: "JevBench v1.4.1 (fstandhartinger/jevbench on GitHub), benchmark for judgment/decision tasks". Below we describe in simple words what benchmarks of this type generally examine.

JevBench measures the abilities the benchmark was built to measure. In detail, it checks:

• tasks representative of the benchmark's stated purpose
• consistent scoring rules so models can be compared fairly
• difficulty calibrated so top models do not max out easily

The score is accuracy: the percentage of tasks answered or solved correctly. 100% means every task was solved perfectly; higher is better.

Official methodology note: JevBench v1.4.1 (fstandhartinger/jevbench on GitHub), benchmark for judgment/decision tasks. Values reported as percentage. Source: https://github.com/fstandhartinger/jevbench

What is the JevBench benchmark?

JevBench is a benchmark for the abilities the benchmark was built to measure. The official description says: 'JevBench v1.4.1 (fstandhartinger/jevbench on GitHub), benchmark for judgment/decision tasks'. Below we describe in simple words what benchmarks of this type generally examine.

What does JevBench test?

JevBench mainly measures the abilities the benchmark was built to measure. Concretely, it checks: tasks representative of the benchmark's stated purpose; consistent scoring rules so models can be compared fairly; difficulty calibrated so top models do not max out easily.

How is the JevBench score calculated?

The score is accuracy: the percentage of tasks answered or solved correctly. 100% means every task was solved perfectly; higher is better. The declared maximum score of this benchmark is 100.0.

Where do I get JevBench data and scores?

The official source is https://github.com/fstandhartinger/jevbench. The official benchmark page/reference is linked on this page. Every score on Open LLM Radar is also available as open data: download the scores JSON (http://open-llm-radar-api:8000/api/v1/benchmarks/jevbench/scores) or pull it directly from the Open LLM Radar API with any HTTP client.