Kindly Request the Inclusion of My Work about llm-sp HOT 3 CLOSED

Thanks for the suggestion! It seems like the approach has two components, but I added it under the guardrail defense section for now. I also added a tldr based on the abstract, but please feel free to provide your own version if you'd like to. It will surely be better than the one I have now :)

from llm-sp.

Thanks for including our work in your list! Your recognition is greatly appreciated.

Regarding the TL;DR section, I would like to propose a slightly revised version for your consideration. I understand that brevity is key, so if you find the revision to be lengthy, you may wish to retain only the first sentence (in bold).

To counter jailbreak attacks, this work proposes a new safety method, Self-Guard, combining the advantages of safety training and safeguards. The method trains the LLM to always append a [harmful] or [harmless] tag to the end of its response before replying to users. In this way, a basic filter can be employed to extract these tags and decide whether to proceed with the response.

I would like to extend my sincere gratitude once again for your dedication and sharing. 😄

from llm-sp.

Of course! Just added the suggested description. Thanks!

from llm-sp.