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

ONNX export from Java #1

Open
Craigacp opened this issue Jan 11, 2022 · 3 comments
Open

ONNX export from Java #1

Craigacp opened this issue Jan 11, 2022 · 3 comments

Comments

@Craigacp
Copy link

I came across this repo from the DJL repo, and you might be interested in the ONNX export functionality we built in Tribuo's last release (v4.2.0). We have a separate module (which only depends on protobuf) that makes writing ONNX models from Java significantly easier, it provides similar helper functionality to the Python ONNX module. We don't have full coverage of the space of ONNX operators yet, but we'd be happy to work with you to expand that set.

The Maven artifact is here, the javadoc is here, and our repo is on GitHub.

You can see examples of how we build ONNX models in the rest of Tribuo (e.g. exporting a FactorizationMachine).

@enpasos
Copy link
Owner

enpasos commented Jan 11, 2022

Thanks for the tip! I'm looking forward to taking a close look.
By the way, do you have a MuZero implementation on Tribuo? ... I am asking because I want to export models trained in DJL from a small MuZero sample implementation to run in a browser using ONNX Runtime Web.

@Craigacp
Copy link
Author

We don't have a MuZero implementation, Tribuo's focus is more on being scitkit-learn in Java than being a general purpose deep learning system. We do classification & regression rather than reinforcement learning, and haven't spent too much time thinking about what an RL system would look like in Java.

@enpasos
Copy link
Owner

enpasos commented Jan 12, 2022

Thanx for the insight to Tribuo.

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