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

Add Flax implementation for BLOOM #17703

Open
2 tasks done
haileyschoelkopf opened this issue Jun 14, 2022 · 7 comments
Open
2 tasks done

Add Flax implementation for BLOOM #17703

haileyschoelkopf opened this issue Jun 14, 2022 · 7 comments

Comments

@haileyschoelkopf
Copy link
Contributor

Model description

I'm interested in adding an implementation of BLOOM in Flax.

The implementation shouldn't be too bad since the pytorch implementation can serve as a guide and a way to check correctness.

Open source status

  • The model implementation is available
  • The model weights are available

Provide useful links for the implementation

@younesbelkada @stas00 @patrickvonplaten

If someone is already planning to work on this then no worries, but if not I will start on this as soon as I have time!

@younesbelkada
Copy link
Contributor

Hi!
Thank you very much for the contribution!
On my side it's a green light since I am not working on it, and it is not on my plans for now. Therefore, I'll be happy to review it! Let us know if you want to work on that :)

@haileyschoelkopf
Copy link
Contributor Author

Thanks! I will open a WIP PR soon and tag you there once I do.

@patrickvonplaten
Copy link
Contributor

Very cool idea - think this can also be a flagship project where we can showcase how to fine-tune BLOOM with Flax cc @patil-suraj @sanchit-gandhi

@patil-suraj
Copy link
Contributor

Awesome! Would be very happy to help with it :)

@sanchit-gandhi
Copy link
Contributor

Great idea! Would also be interested in getting involved, this would be a super cool model addition!

@haileyschoelkopf
Copy link
Contributor Author

Thanks everyone for the interest! I'd love to collaborate with you all.

I'm hoping to push a rough draft of modeling code by the end of the weekend (earlier if I have time), and will tag you all when I open the PR with that. Does that sound alright?

@haileyschoelkopf
Copy link
Contributor Author

I've opened a PR (and documented the state of the in-progress code I'm still working on) at #17761 ! We can discuss further in that PR how to collaborate / proceed.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

5 participants