Dilshod Tadjibaev

RTMW3D-x is buildable with https://github.com/tracel-ai/burn/pull/3564 (still under review)

So much has changed, we should probably create a new PR by now.

CI is failing

I think we should make "an attribute of the model struct and declared during Model::new()". Record loading happens at the module level and we plan to support multi/sub graphs as...

Regarding testing. Yeah I wish we could store legacy ONNX files but there are a few issues: 1) License and size (which might slow down testing). However, we can probably...

I have reviewed the PR and believe a simpler and more scalable solution would be to extract inputs with initializers for node types such as `Sub`, `Add`, etc., as `Constants`....

`keepdims == 0` has been fixed by https://github.com/tracel-ai/burn/pull/3497

If you're inquiring about the compatibility of Burn with the StableHLO serialization, then Burn is well equipped for the task. At present, Burn facilitates the importation of ONNX models, encompassing...

@crutcher, might have missed a notification. Still valid?

Ignored tests just in case the code gets deleted: ```rust #[test] #[ignore = "https://github.com/tracel-ai/burn/issues/2080"] fn test_1d_bicubic() { // Initialize the model without weights (because the exported file does not contain...