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

[CUDA] Objective functions and metrics for cuda_exp #5163

Open
Tracked by #5153
shiyu1994 opened this issue Apr 21, 2022 · 2 comments
Open
Tracked by #5153

[CUDA] Objective functions and metrics for cuda_exp #5163

shiyu1994 opened this issue Apr 21, 2022 · 2 comments

Comments

@shiyu1994
Copy link
Collaborator

Summary

The cuda_exp in current master branch includes only a new tree learner version for CUDA. Objectives (boosting) and metrics (evaluation) should be implemented on CUDA.

@shiyu1994 shiyu1994 mentioned this issue Apr 21, 2022
60 tasks
@jameslamb
Copy link
Collaborator

Thanks for writing this up @shiyu1994 !

Can you add some more details here, to help others from the community who might want to help contribute to this?

For example...would it be possible to implement metrics one at a time (e.g. a PR like "add CUDA implementation of MAPE") and have LightGBM choose whether to evaluate it on the GPU or CPU? Or would this require a single PR implementing all metrics?

@shiyu1994
Copy link
Collaborator Author

For example...would it be possible to implement metrics one at a time

Sure. That's what I have planned. Each PR should contain only one or several related objectives/metrics, to keep the PRs as small as possible.

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

2 participants