Phi-1.5
开源来自 Microsoft · 发布于 2024-01-01
16.3
平均分
N/A
输入价格
N/A
输出价格
N/A
上下文窗口
text
类型
Tested on 11 benchmarks with 16.3% average. Top scores: Winogrande (46.8%), HellaSwag (30.1%), ARC AI2 (25.9%).
基准测试分数
| 基准测试 | 类别 | 分数 | Bar |
|---|---|---|---|
| Winogrande | knowledge | 46.8 | |
| HellaSwag | knowledge | 30.1 | |
| ARC AI2 | knowledge | 25.9 | |
| IFEval | language | 20.3 | |
| MMLU | knowledge | 16.8 | |
| OpenBookQA | knowledge | 16.3 | |
| MMLU-PRO | knowledge | 7.7 | |
| BBH (HuggingFace) | general | 7.5 | |
| MUSR | reasoning | 3.4 | |
| GPQA | knowledge | 2.4 | |
| MATH Level 5 | math | 1.8 |