最近在搞 n8n 工作流，结合 mcp playwright ，做网站监控
但是消耗的 token 太多了，网站网页很多冗余的代码，一下就几万的 token 没了
所以想着还是用下本地模型
This topic created in 443 days ago, the information mentioned may be changed or developed.
 |
S1ahs3rAug 20, 2025
1
用 modelfile 形式，改 template 。
https://ollama.com/library/qwen3:0.6b/blobs/ae370d884f10 |
 |
taco1wangAug 21, 2025
2
前面加个 <no_think> 就行。。
|
 |
cogearAug 21, 2025
5
|
 |
lchynnAug 27, 2025
6
提示词的最后加上（换行） /no_think 即可。
|
|
lchynnAug 27, 2025
7
Specifically, you can add /think and /no_think to user prompts or system messages to switch the model's thinking mode from turn to turn. The model will follow the most recent instruction in multi-turn conversations.
https://huggingface.co/Qwen/Qwen3-8B |
Select Language