测试版
排行榜/Qwen2.5 32B Instruct
Alibaba logo

Qwen2.5 32B Instruct

开源

来自 Alibaba · 发布于 2024-09-17

43.2
平均分
N/A
输入价格
N/A
输出价格
N/A
上下文窗口
text-generation
类型

Tested on 7 benchmarks with 43.2% average. Top scores: IFEval (83.5%), MATH Level 5 (62.5%), BBH (HuggingFace) (56.5%).

基准测试类别分数Bar
IFEvallanguage83.5
MATH Level 5math62.5
BBH (HuggingFace)general56.5
MMLU-PROknowledge51.9
PropensityBenchsafety22.9
MUSRreasoning13.5
GPQAknowledge11.7