Ke Zhang

There's one more scenario happened frequently is, do some manual constant folding optimization. Say, it's known that a sub-graph can be constant-folded, and user would like to directly remove the...

Thanks! PRs are welcome to add more necessary types' support for existing ops.

are you interested in proposing op definitions PRs needed for it?

I agree. That means for optimization like "constant folding" should not be part of ONNX repos.

Quick question: for TF and PyTorch users, do they need to follow different APIs to feed data in EDL? @QiJune

+1. This may make ONNX to be usable in many recommendation models.

+@ebarsoum who's running operator sig. @yaoyuannnn you may also want to send a PR to propose such an operator or function please.

@yaoyuannnn any progress on contributing an embedding function please?

Thank you! @edgchen1 Yep. this looks confusing... @postrational any comments on this? I'd suggest to deprecate one mechanism of storing large tensors if possible.

@houseroad would you help to fix the variable name issue for models that @youngkim93 mentioned above please? Thank you!