phi-3-small 7.4B
开源来自 Microsoft · 发布于 2024-01-01
67.4
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 8 benchmarks with 67.4% average. Top scores: ARC AI2 (87.6%), OpenBookQA (84.0%), BBH (72.1%).
基准测试分数
| 基准测试 | 类别 | 分数 | Bar |
|---|---|---|---|
| ARC AI2 | knowledge | 87.6 | |
| OpenBookQA | knowledge | 84.0 | |
| BBH | reasoning | 72.1 | |
| HellaSwag | knowledge | 69.3 | |
| MMLU | knowledge | 67.6 | |
| Winogrande | knowledge | 63.0 | |
| TriviaQA | knowledge | 58.1 | |
| ANLI | knowledge | 37.1 |