Llama 3.1 8B Instruct
开源来自 Meta · 发布于 2024-07-23
27.4
平均分
$0.02/1M
输入价格
$0.05/1M
输出价格
16K tokens (~8 books)
上下文窗口
text
类型
Tested on 16 benchmarks with 27.4% average. Top scores: Chatbot Arena Elo — Overall (1211.0%), GSM8K (82.4%), PIQA (62.4%).
基准测试分数
| 基准测试 | 类别 | 分数 | Bar |
|---|---|---|---|
| Chatbot Arena Elo — Overall | arena | 1211.0 | |
| GSM8K | math | 82.4 | |
| PIQA | knowledge | 62.4 | |
| IFEval | language | 50.6 | |
| MMLU | knowledge | 41.5 | |
| Aider — Code Editing | coding | 37.6 | |
| MMLU-PRO | knowledge | 30.9 | |
| BBH (HuggingFace) | general | 29.2 | |
| MATH level 5 | math | 22.9 | |
| MATH Level 5 | math | 15.5 | |
| Balrog | knowledge | 15.1 | |
| GPQA | knowledge | 9.5 | |
| MUSR | reasoning | 8.5 | |
| OTIS Mock AIME 2024-2025 | math | 2.4 | |
| WeirdML | coding | 1.7 | |
| GPQA diamond | knowledge | 1.3 |