-
Notifications
You must be signed in to change notification settings - Fork 102
Add interactive generation script #43
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
Conversation
Co-authored-by: Thomas Wang <24695242+thomasw21@users.noreply.github.com> Co-authored-by: Narsil <Narsil@users.noreply.github.com>
- remove eval - remove pipeline & json from import
|
Maybe I'm stupid, but it doesn't seem to use accelerate? |
|
Using |
|
Then this flag will automatically distribute the model into the available GPUs only if accelerate is installed, and do the offloading |
|
Nice. Nothing builds intuition for the models like interactive mode. |
thomasw21
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Haven't had the time to test it out myself, but I'm pretty sure you've tested things yourselves @younesbelkada, so approving. @Muennighoff might have something to say before merging :D
Co-authored-by: Niklas Muennighoff <n.muennighoff@gmail.com>
Co-authored-by: Niklas Muennighoff <n.muennighoff@gmail.com>
|
LGTM 🚀 |
Muennighoff
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
lgtm; nice!
I added an interactive generation script that works fine with the latest release of
transformersTested on a JZ node and seems to work fine!
Based on a combination of @thomasw21 and @Narsil scripts
This code needs
accelerateto be installed first in order to run properly