Model

LLaMA 13B

Meta · open weights

Knowledge & Reasoning

Benchmark	Score	Metric	Harness	Source	Confidence	Link
ARC-Challenge	52.7	accuracy	N/A	Independent	medium	arxiv.org
ARC-Easy	74.8	accuracy	N/A	Independent	medium	arxiv.org
BBH	37.9	average	N/A	Independent	medium	arxiv.org
BBH	37.0	average	N/A	Self-reported	medium	arxiv.org
BoolQ	78.7	score	N/A	Independent	medium	arxiv.org
BoolQ	78.1	accuracy	N/A	Independent	medium	arxiv.org
BoolQ	78.1	score	N/A	Self-reported	medium	arxiv.org
HellaSwag	79.2	accuracy	N/A	Independent	medium	arxiv.org
HellaSwag	79.2	overall_accuracy	N/A	Self-reported	medium	arxiv.org
LAMBADA	75.2	score	N/A	Independent	medium	arxiv.org
MMLU	47.7	exact_match	N/A	Independent	medium	arxiv.org
MMLU	46.9	exact_match	N/A	Self-reported	medium	arxiv.org
OpenBookQA	56.4	accuracy	N/A	Self-reported	medium	arxiv.org
OpenBookQA	56.4	accuracy	N/A	Independent	medium	arxiv.org
PIQA	80.1	accuracy	N/A	Independent	medium	arxiv.org
PIQA	80.1	score	N/A	Self-reported	medium	arxiv.org
PIQA	80.1	score	N/A	Independent	medium	arxiv.org
ScienceQA	43.3	score	N/A	Independent	medium	scienceqa.github.io
siqa	50.4	accuracy	N/A	Independent	medium	arxiv.org
Trivia QA	77.9	exact_match	N/A	Self-reported	medium	arxiv.org
WinoGrande	73.0	accuracy	N/A	Independent	medium	arxiv.org
WinoGrande	73.0	accuracy	N/A	Self-reported	medium	arxiv.org

Benchmark	Score	Metric	Harness	Source	Confidence	Link
GSM8K	20.6	exact_match	N/A	Independent	medium	arxiv.org
GSM8K	17.8	accuracy	N/A	Independent	medium	arxiv.org
GSM8K	17.8	exact_match	N/A	Self-reported	medium	arxiv.org
MATH	3.9	accuracy	N/A	Independent	medium	arxiv.org

Benchmark	Score	Metric	Harness	Source	Confidence	Link
MBPP	22.0	accuracy	N/A	Independent	medium	arxiv.org