Model

PaLM 2-M

Google

deepmind.google

Knowledge & Reasoning

Knowledge & Reasoning
BenchmarkScoreMetricHarnessSourceConfidenceLink
ARC-Challenge64.9accuracyN/AIndependentmediumarxiv.org
ARC-Easy88.0accuracyN/AIndependentmediumarxiv.org
BoolQ88.6accuracyN/AIndependentmediumarxiv.org
BoolQ88.6scoreN/AIndependentmediumarxiv.org
CommitmentBank80.4accuracyN/AIndependentmediumarxiv.org
HellaSwag86.7accuracyN/AIndependentmediumarxiv.org
HellaSwag84.0overall_accuracyN/ASelf-reportedmediumarxiv.org
LAMBADA83.7accuracyN/AIndependentmediumarxiv.org
MultiRC84.1f1N/AIndependentmediumarxiv.org
Natural Questions32.0exact_matchN/AIndependentmediumarxiv.org
OpenBookQA57.4accuracyN/ASelf-reportedmediumarxiv.org
OpenBookQA56.2accuracyN/AIndependentmediumarxiv.org
PIQA83.2accuracyN/AIndependentmediumarxiv.org
RTE81.9accuracyN/AIndependentmediumarxiv.org
StoryCloze86.7accuracyN/AIndependentmediumarxiv.org
Trivia QA81.7exact_matchN/AIndependentmediumarxiv.org
Trivia QA81.7exact_matchN/ASelf-reportedmediumarxiv.org
WebQuestions26.9exact_matchN/AIndependentmediumarxiv.org
WiC52.0accuracyN/AIndependentmediumarxiv.org
WinoGrande79.2accuracyN/AIndependentmediumarxiv.org
WinoGrande79.2accuracyN/ASelf-reportedmediumarxiv.org
WSC88.1accuracyN/AIndependentmediumarxiv.org

Multimodal

Multimodal
BenchmarkScoreMetricHarnessSourceConfidenceLink
COPA90.0accuracyN/AIndependentmediumarxiv.org

Reproduced by the community

Loading community runs…