qwen2.5:0.5b-base-q4_0 - Ollama 框架

它拥有**显著更多的知识**，并且在 **编码** 和 **数学** 方面的能力大大增强，这归功于这些领域的专业专家模型。
它在 **指令遵循**、**长文本生成**（超过 8K 个 tokens）、**理解结构化数据**（例如，表格）和 **生成结构化输出** 方面取得了重大进展，尤其是在 JSON 格式中。它也**更能适应各种系统提示**，从而改善了聊天机器人的角色扮演和条件设置。
它支持高达 128K 个 tokens 的 **长上下文**，并且可以生成多达 8K 个 tokens。
它为超过 29 种语言提供 **多语言支持**，包括中文、英语、法语、西班牙语、葡萄牙语、德语、意大利语、俄语、日语、韩语、越南语、泰语、阿拉伯语等。

请注意：除 3B 和 72B 型号外，所有型号均在 Apache 2.0 许可下发布，而 3B 和 72B 型号则在 Qwen 许可下发布。

参考资料

GitHub

博客文章

HuggingFace

Qwen2.5 is the latest series of Qwen large language models. For Qwen2.5, a range of base language models and instruction-tuned models are released, with sizes ranging from 0.5 to 72 billion parameters. Qwen2.5 introduces the following improvements over Qwen2:

- It possesses **significantly more knowledge** and has greatly enhanced capabilities in **coding** and **mathematics**, due to specialized expert models in these domains.
- It demonstrates significant advancements in **instruction following**, **long-text generation** (over 8K tokens), **understanding structured data** (e.g., tables), and **generating structured outputs**, especially in JSON format. It is also **more resilient to diverse system prompts**, improving role-play and condition-setting for chatbots.
- It supports **long contexts** of up to 128K tokens and can generate up to 8K tokens.
- It offers **multilingual support** for over 29 languages, including Chinese, English, French, Spanish, Portuguese, German, Italian, Russian, Japanese, Korean, Vietnamese, Thai, Arabic, and more.

Please note: all models except the 3B and 72B are released under the Apache 2.0 license, while the 3B and 72B models are under the Qwen license.

## References

[GitHub](https://github.com/QwenLM/Qwen2.5)

[Blog post](https://qwenlm.github.io/blog/qwen2.5/)

[HuggingFace](https://hugging-face.cn/collections/Qwen/qwen25-66e81a666513e518adb90d9e)

粘贴、拖放或单击以上传图片 (.png, .jpeg, .jpg, .svg, .gif)