Replace Fbgemm related operators with Pytorch aten operators

[majing921201]

hi,
I found that Pytorch started to support fbgemm operators: https://github.com/pytorch/pytorch/blob/2cc01cc6d3ad2aff47e8460667ba654b2e4c9f21/aten/src/ATen/native/native_functions.yaml#L14784 .
The kernels implementation in PyTorch are totally the same as Fbgemm.
Would you like to switch all of the torch.ops.fbgemm.xxxx usage in model to torch.ops.aten.xxx, for torch.xxx() in future ?

[jiaqizhai]

Hi - thanks for your interest in our work!

We do utilize many jagged operators that are not in fbgemm (see e.g., https://github.com/facebookresearch/generative-recommenders/blob/main/generative_recommenders/ops/triton/triton_jagged.py) so this may not be high priority on our radar yet. We can consolidate as we further refactor codebase in the future (Xing/Linjian can probably comment more on this).

[majing921201]

Thanks for your reply. I will follow up codebase change.