Glossary

Discover ARC Benchmark: AI2 Reasoning Challenge includes two sets of questions: ARC‑e & ARC‑c. Evaluate reasoning knowledge.

Discover PIQA, a benchmark that evaluates a model’s comprehension of physical interactions. It challenges models to choose appropriate actions in various scenarios. Sources included.

Discover the Winogrande dataset, designed for AI model training and testing in common sense reasoning. Evaluate AI’s language understanding with its challenging problems and expanded scale.

Discover how the HellaSwag benchmark assesses a model’s common sense reasoning ability by presenting scenarios that test implicit knowledge about the world.