Phi 2
开源来自 Microsoft · 发布于 2023-12-13
30.2
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 14 benchmarks with 30.2% average. Top scores: ARC AI2 (67.9%), OpenBookQA (64.8%), BBH (45.9%).
基准测试分数
| 基准测试 | 类别 | 分数 | Bar |
|---|---|---|---|
| ARC AI2 | knowledge | 67.9 | |
| OpenBookQA | knowledge | 64.8 | |
| BBH | reasoning | 45.9 | |
| TriviaQA | knowledge | 45.2 | |
| MMLU | knowledge | 44.5 | |
| HellaSwag | knowledge | 38.1 | |
| BBH (HuggingFace) | general | 28.0 | |
| IFEval | language | 27.4 | |
| MMLU-PRO | knowledge | 18.1 | |
| MUSR | reasoning | 13.8 | |
| ANLI | knowledge | 13.8 | |
| Winogrande | knowledge | 9.4 | |
| MATH Level 5 | math | 3.0 | |
| GPQA | knowledge | 2.9 |