Skip to content
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

[Feature] 建议部署时增加环境变量,配置是否注入系统级提示消息 #3665

Closed
Scorpio-xu opened this issue Dec 27, 2023 · 12 comments

Comments

@Scorpio-xu
Copy link

Scorpio-xu commented Dec 27, 2023

建议部署时增加环境变量,配置是否注入系统级提示消息
我使用的是 Gemini 模型,分享给别人后,别人老是说我这是 ChatGPT,是因为系统发了一条“you are chatgpt……”的消息,然后我就得告诉他在设置中把“注入系统级提示消息”这个设置项关掉。
能不能把这个设置项改到环境变量里啊,或者默认关闭,谢谢大佬

@Issues-translate-bot
Copy link

Bot detected the issue body's language is not English, translate it automatically.


Title: [Feature] It is recommended to add environment variables during deployment and configure whether to inject system-level prompt messages

It is recommended to add environment variables during deployment and configure whether to inject system-level prompt messages
I use the Gemini model. After sharing it with others, they always say that this is ChatGPT because the system sends a message "you are chatgpt...", and then I have to tell them to inject the system level in the settings. The prompt message is turned off.
Can you change this setting item to an environment variable, or turn it off by default? Thank you, boss

@H0llyW00dzZ
Copy link
Contributor

works fine now for me

image

been working on this
H0llyW00dzZ@b9ed661

@fredliang44
Copy link
Collaborator

@Scorpio-xu injecting system message is useful in most cases, I guess we might need to support different system default message in the future to adjust different model provider

@H0llyW00dzZ
Copy link
Contributor

H0llyW00dzZ commented Dec 27, 2023

@Scorpio-xu injecting system message is useful in most cases, I guess we might need to support different system default message in the future to adjust different model provider

yeah
not right now, gemini still unstable I've been testing on it

@walterclozet
Copy link

@Scorpio-xu injecting system message is useful in most cases, I guess we might need to support different system default message in the future to adjust different model provider

can i set these system messages via environment variables?

@Scorpio-xu
Copy link
Author

Scorpio-xu commented Dec 28, 2023

works fine now for me

image

been working on this H0llyW00dzZ@b9ed661

@H0llyW00dzZ Thank you for your image. The image falsely claims to be ChatGPT, but in reality, it is not.

@Scorpio-xu
Copy link
Author

Scorpio-xu commented Dec 28, 2023

@Scorpio-xu injecting system message is useful in most cases, I guess we might need to support different system default message in the future to adjust different model provider

@fredliang44 Yes, it is indeed useful, but I didn't say to turn it off. Instead, I suggested adding an environment variable to control it, allowing the service provider to decide whether to enable it. After adding this sentence, he claimed to be ChatGPT, but he clearly is not, causing a misunderstanding.

@PaperPlaneDeemo
Copy link

i have the same issue, when i use gemini pro, the system injects system-level prompt messages that says you are chatgpt. How can I turn it off when i use gemini pro?

@H0llyW00dzZ
Copy link
Contributor

i have the same issue, when i use gemini pro, the system injects system-level prompt messages that says you are chatgpt. How can I turn it off when i use gemini pro?

in chat settings

image

@QAbot-zh
Copy link

同意,至少应该默认为不注入

@Issues-translate-bot
Copy link

Bot detected the issue body's language is not English, translate it automatically.


Agree, at least the default should be no injection

@magicds
Copy link

magicds commented Nov 24, 2024

I believe this design is quite poor, and it has been set as the default behavior. I am perplexed; can I disable it globally?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

8 participants