Llama 7B — Benchmarks

Benchmark scores for Llama 7B aggregated from public leaderboards, with how it ranks among open models. See hardware requirements for what you need to run it.

Overall rank: #86 of 123 open modelscomposite 37.1/100 across 10 benchmarks in 3 categories · methodology

Knowledge

BenchmarkScoreOpen rankAll models
HellaSwag76.2%#23 / 42#43 / 76
ARC Challenge47.6%#32 / 51#50 / 77
BoolQ76.5%#18 / 33#46 / 77
OpenBookQA57.2%#11 / 21#20 / 42
TriviaQA71.0%#12 / 18#30 / 39
MMLU35.6%#72 / 77#126 / 136

Math

BenchmarkScoreOpen rankAll models
GSM8K11.0%#56 / 60#79 / 93

Reasoning

BenchmarkScoreOpen rankAll models
BIG-Bench Hard33.5%#34 / 37#47 / 50
WinoGrande70.1%#25 / 46#51 / 80
PIQA79.8%#21 / 35#39 / 60

Scores aggregated from public benchmark sources (each linked from the benchmark pages). llmrun does not run these benchmarks.