II.
Benchmark overview
Reference · livebenchmark:bbh
BIG-Bench Hard (BBH) overview
23 challenging tasks from BIG-Bench where prior LLMs underperformed average human raters; widely used as a model reasoning benchmark.
Attributes
displayName
BIG-Bench Hard (BBH)
homepageUrl
kind
reasoning
targetsKind
ModelVersion
description
23 challenging tasks from BIG-Bench where prior LLMs underperformed
average human raters; widely used as a model reasoning benchmark.
Outgoing edges
applies_to1
- domain:computer-science·DomainComputer Science
covers1
- skill-area:general-knowledge-reasoning·SkillAreaGeneral Knowledge Reasoning
Incoming edges
None.