-
Notifications
You must be signed in to change notification settings - Fork 476
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
请教一下关于工具调用的微调数据 #684
Comments
assistant是大模型的回复,observation是工具返回的结果。
|
感谢指教,过几天尝试一下~ |
@zRzRzRzRzRzRzR 您好,想再次请教一下,toolcall微调每条数据只能有一个tool吗,还是可以写进多个tool,比如:
因为我发现,如果工具列表有多个工具,传递给engine.generate的input参数是:
所以不知道微调数据是要:
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Feature request / 功能建议
看到https://zhipu-ai.feishu.cn/wiki/UOmbw8PsJiJXrfksUmwcnHHBnQf上有给出tools调用的微调模板,想请教一下关于工具调用的微调数据:
例如我现在只有"weather"工具,数据集格式如上。
上面这俩role是为了让微调后模型根据user的输入而输出固定的格式化参数。我看上面文档除了这俩还有"role": "observation"和 "role": "assistant",这两个作用是什么?
假如我得到了第一轮输出的工具参数
weather\n{"city": "深圳"}
,参数传入weather工具,是不是应该直接将工具返回的结果:作为LLM下一轮对话的"role": "system"的内容?
然后微调数据变成:
不知道理解的对不对,还望解答。
Motivation / 动机
Your contribution / 您的贡献
The text was updated successfully, but these errors were encountered: