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

Difference with speculative #1

Open
WilliamsToTo opened this issue Feb 24, 2025 · 1 comment
Open

Difference with speculative #1

WilliamsToTo opened this issue Feb 24, 2025 · 1 comment

Comments

@WilliamsToTo
Copy link

What's the difference between CleanGen and speculative decoding baseline? I think CleanGen is very similar to speculative decoding. The main difference is that two models in CleanGen have the same size but speculative decoding requires a smaller model and a larger model.

@1GaryLi
Copy link
Collaborator

1GaryLi commented Feb 24, 2025

Thank you for your question. The main difference is that CleanGen has different algorithms to replace a "suspicious" token during decoding. Speculative's main goal is that the decoding output should always follow the distribution of the target model (thus if the target model is backdoored, the output is still compromised). CleanGen's main goal is that when the input contains trigger, the output should follow the distribution of reference model, when the input has no trigger, the output should follow the target model. Hope this will help your question!

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