HF Model & Benchmark Matrix

24,078 variants · 14,984 scores · 120 benchmarks
Home/Orgs/Qwen/Qwen/Qwen2-7B

Qwen/Qwen2-7B

family qwen · 6 variants · unknown

provider
Qwen
released
—
huggingface
https://huggingface.co/Qwen/Qwen2-7B
best scores
MMLU: 68.4MMLU-Pro: 41.0GPQA: 34.2MT-Bench: 8.05HumanEval: 62.2MBPP: 67.9EvalPlus: 60.9LiveCodeBench: 17.3GSM8K: 79.6MATH: 30.0C-Eval: 47.4BBH: 56.1HellaSwag: 83.2WinoGrande: 78.4TruthfulQA: 42.2CMMLU: 50.8

Best scores per benchmark

benchmarkcategorybest score
HellaSwagknowledge83.2
GSM8Kmath79.6
WinoGrandeknowledge78.4
MMLUknowledge68.4
MBPPcoding67.9
HumanEvalcoding62.2
EvalPluscoding60.9
BBHreasoning56.1
CMMLUknowledge50.8
C-Evalknowledge47.4
TruthfulQAsafety42.2
MMLU-Proknowledge41.0
GPQAreasoning34.2
MATHmath30.0
LiveCodeBenchcoding17.3
MT-Benchchat8.05

Variants (6)

variantkindquantbitsvRAMdlorgyear
Qwen2-7B-Instructoriginalbf1616.0629,607Qwen2024
Qwen2-7Boriginalbf1616.089,240Qwen2024
Qwen2-7B-Instruct-bnb-4bitbnb-4bit4bit1,993unsloth2024
Samantha-Qwen-2-7B-GGUFggufgguf2.0554bartowski2024
Qwen2-7B-bnb-4bitbnb-4bit4bit469unsloth2024
Qwen2-7B-Instruct-MLXmlxmlx57Qwen2024
Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.