Skip to content

fix skip unnecessary check of gram_matrix in coordinate_descent #22008

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

Closed
wants to merge 1 commit into from

Conversation

derHeinzer
Copy link

Example: Fixes #21997. See also #19004.

@agramfort
Copy link
Member

@derHeinzer can you add a test to better understand what you fix?

@QuantHao
Copy link

Hi, I encounter the same problem as #21997, thus I curious when this fix could be published? Thanks.

@valkmit
Copy link

valkmit commented Jan 9, 2023

This is a problem still using LassoCV w/ float32 data, would appreciate this getting into master!

@jeremiedbb
Copy link
Member

@valkmit, the fix for issue #21997 has been released in v1.2. Could you try with the latest version of scikit-learn ? If you still find issues with LassoCV on float32 data, can you provide some reproducible code ?

@jeremiedbb jeremiedbb added the Needs Reproducible Code Issue requires reproducible code label Feb 28, 2023
@valkmit
Copy link

valkmit commented Mar 23, 2023

@jeremiedbb Thanks for the follow up! Unfortunately I ended up switching to float64 and changing how I sampled a while back to address the issue and no longer have the data needed to generate this issue (it was a very large dataset)

@jeremiedbb
Copy link
Member

I'm closing this PR since the issue it is supposed to fix has already been fixed, assuming this one is now unecessary. If not, an issue with some reproducible code will be welcome before reopening it.

@jeremiedbb jeremiedbb closed this Mar 7, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
module:linear_model Needs Reproducible Code Issue requires reproducible code
Projects
None yet
Development

Successfully merging this pull request may close these issues.

validation check for precomputed gram matrix fails erroneously when using float32 data
5 participants