信息
“智闻AI“ 是由人工智能编撰的刊物集合，确保您只获得最有价值的信息，旨在助您消除信息差，突破信息茧房的局限。 了解更多 >>
Qwen-72B：大型语言与多模态模型的一次突破
- summary
- score
Qwen2，一系列新型大型语言和多模态模型，超越了 其前代及竞争对手。核心模型Qwen2-72B在多项测试中表现优异：MMLU得分84.2，GPQA得分37.9，HumanEval得分64.6，GSM8K得分89.5，BBH得分82.4。该模型支持30种语言，从英语到中文。
模型权重及资源在Hugging Face、ModelScope和GitHub上免费提供，支持定制与部署。
| Scores | Value | Explanation |
|---|---|---|
| Objectivity | 7 | Comprehensive, balanced reporting with in-depth analysis. |
| Social Impact | 5 | Significant influence on public opinion in tech communities. |
| Credibility | 6 | Verified by multiple sources, highly credible. |
| Potential | 6 | High potential for significant tech advancements. |
| Practicality | 5 | Widely applied in practice, achieving good results. |
| Entertainment Value | 3 | Somewhat entertaining, primarily for tech enthusiasts. |