U
INTELLECT-1
来自 Unknown · 发布于 2024-01-01
20.2
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 12 benchmarks with 20.2% average. Top scores: HellaSwag (61.9%), ARC AI2 (39.4%), GSM8K (38.6%).
基准测试分数
| 基准测试 | 类别 | 分数 | Bar |
|---|---|---|---|
| HellaSwag | knowledge | 61.9 | |
| ARC AI2 | knowledge | 39.4 | |
| GSM8K | math | 38.6 | |
| MMLU | knowledge | 33.2 | |
| Winogrande | knowledge | 31.6 | |
| IFEval | language | 17.6 | |
| BBH | reasoning | 13.1 | |
| MUSR | reasoning | 4.1 | |
| MMLU-PRO | knowledge | 1.3 | |
| BBH (HuggingFace) | general | 1.0 | |
| MATH Level 5 | math | 0.0 | |
| GPQA | knowledge | 0.0 |