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

Some failed prompt but caused a jailbreak #7

Open
zichuan-liu opened this issue Jan 12, 2024 · 3 comments
Open

Some failed prompt but caused a jailbreak #7

zichuan-liu opened this issue Jan 12, 2024 · 3 comments

Comments

@zichuan-liu
Copy link

zichuan-liu commented Jan 12, 2024

In the samples, there are some generated prompts that turn into [new prompt], resulting in a jailbreak message when calculating the result. The probability of it appearing is about 1 in 20.
e.g,
1705051804857

@zhu111-eng
Copy link

![Uploading 1RF1]1X9}VE]Q}R2SYOD2.png…]()

In the samples, there are some generated prompts that turn into [new prompt], resulting in a jailbreak message when calculating the result. The probability of it appearing is about 1 in 20. e.g, 1705051804857

Hello, I encountered this problem during the reproduction process, consulted some information, and also tried to modify the code, but the limited ability is still not solved, so I would like to ask you how to solve it~
QQ图片20240122091830

@zichuan-liu
Copy link
Author

![Uploading 1RF1]1X9}VE]Q}R2SYOD2.png…]()

In the samples, there are some generated prompts that turn into [new prompt], resulting in a jailbreak message when calculating the result. The probability of it appearing is about 1 in 20. e.g, 1705051804857

Hello, I encountered this problem during the reproduction process, consulted some information, and also tried to modify the code, but the limited ability is still not solved, so I would like to ask you how to solve it~ QQ图片20240122091830

that's same ur conv has no "set_sys_msg", check out ur fastchat version it may have correct template "from fastchat.model import get_conversation_template"

@zhu111-eng
Copy link

![Uploading 1RF1]1X9}VE]Q}R2SYOD2.png…]()

In the samples, there are some generated prompts that turn into [new prompt], resulting in a jailbreak message when calculating the result. The probability of it appearing is about 1 in 20. e.g, 1705051804857

Hello, I encountered this problem during the reproduction process, consulted some information, and also tried to modify the code, but the limited ability is still not solved, so I would like to ask you how to solve it~ QQ图片20240122091830

that's same ur conv has no "set_sys_msg", check out ur fastchat version it may have correct template "from fastchat.model import get_conversation_template"

Thank you for your reply.

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

2 participants