qwen2.5:0.5b-base-q4_0

由於這些領域的專業模型，它擁有顯著更多的知識，並且在程式碼編寫和數學方面的能力大大增強。
它在指令遵循、長文本生成（超過 8K tokens）、理解結構化資料（例如，表格）和生成結構化輸出（尤其是在 JSON 格式中）方面展現了顯著的進步。它也更能適應多樣化的系統提示，從而改善了聊天機器人的角色扮演和條件設定。
它支援高達 128K tokens 的長上下文，並且可以生成高達 8K tokens。
它為超過 29 種語言提供多語言支援，包括中文、英文、法文、西班牙文、葡萄牙文、德文、義大利文、俄文、日文、韓文、越南文、泰文、阿拉伯文等等。

請注意：除了 3B 和 72B 模型之外，所有模型均根據 Apache 2.0 授權條款發布，而 3B 和 72B 模型則根據 Qwen 授權條款發布。

參考資料

GitHub

部落格文章

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://huggingface.co/collections/Qwen/qwen25-66e81a666513e518adb90d9e)

貼上、拖放或點擊以上傳圖片 (.png, .jpeg, .jpg, .svg, .gif)