How to put the thinking process into thinking in DeepSeek R1 deployment in xinference

[ghw-bit]

Self Checks

• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
• [FOR CHINESE USERS] 请务必使用英文提交 Issue，否则会被关闭。谢谢！:）
• Please do not modify this template :) and fill in all the required fields.

1. Is this request related to a challenge you're experiencing? Tell me about your story.

How can I use the DeepSeek R1 deployed in Xinference to output the thinking process in Thinking instead of the normal output? Currently, only the official interface and the deepseek-reasoner model can put the thinking process in Thinking.

2. Additional context or comments

No response

[Mercure28]

To enable DeepSeek R1 deployed in Xinference to output the thinking process in "Thinking" instead of the normal output, you may need to modify the model's response behavior through prompt engineering or adjust the underlying inference settings. Since only the official interface and the deepseek-reasoner model currently support this feature, consider integrating DeepSeek Reasoner with your Xinference deployment or using a system prompt that explicitly instructs the model to verbalize its reasoning steps. Additionally, check if Xinference allows fine-tuning or configuring intermediate reasoning outputs, as some frameworks offer customizable response formats. If the feature is restricted by the model itself, you might need to use an external wrapper or middleware to structure and display the reasoning process explicitly.