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

Is it possible/recommended to do open generation? #10

Closed
LorrinWWW opened this issue Aug 17, 2022 · 4 comments
Closed

Is it possible/recommended to do open generation? #10

LorrinWWW opened this issue Aug 17, 2022 · 4 comments

Comments

@LorrinWWW
Copy link

Thanks for sharing the great work!

I'm curious if it is appropriate to do open generation. E.g. put only [gMASK] at the beginning and then complete the text.

Thank you again!

@Sengxian
Copy link
Contributor

Sengxian commented Aug 17, 2022

Hello @LorrinWWW. GLM-130B can still generate fluent sentences in this scenario, you can also generate different sentences by changing the seed. However, we are not sure if it is appropriate, it maybe depends on what you want to do.

@LorrinWWW
Copy link
Author

Thanks for your answer! Another question is do you have any suggestion on performing batch generation?

@Sengxian
Copy link
Contributor

Currently we don't support batch generation, but I'm guessing it's not too hard to modify the autoregressive generation logic and its corresponding strategy in SAT to support batch generation with a fixed context length (or no context).

@LorrinWWW
Copy link
Author

Thank you very much!

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