torch.aten.reflection_pad1d

[kumardeepakamd]

Support torch.aten.reflection_pad1d in torch MLIR

[frederik-h]

I am working on the torch.aten.reflection_pad2d operator and it is essentially done, but I won't be able to finish the PR before Christmas. I am not sure if it is worth to do a separate implementation of the 1d operator, i.e. it should be possible to reuse and simplify the 2d implementation.

[stellaraccident]

@kumardeepakamd / @frederik-h can you coordinate?

[frederik-h]

@kumardeepakamd / @frederik-h can you coordinate?

@kumardeepakamd I have created a draft PR for the 2d operator: llvm/torch-mlir#2693. I think the 1d operator is just a special case, i.e. it should probably be converted to the 2d operator, but it will still require separate testing etc.

[kumardeepakamd]

[AMD Official Use Only - General] I am thinking I still take care of pad1d and will make use of pad2d work by @frederik<https://github.com/frederik-h>. Is that okay? Sent from my T-Mobile 5G Device Get Outlook for Android<https://aka.ms/AAb9ysg>

…

________________________________ From: Frederik Harwath ***@***.***> Sent: Friday, December 22, 2023 6:46:25 AM To: nod-ai/SHARK-Turbine ***@***.***> Cc: Deepak, Kumar ***@***.***>; Mention ***@***.***> Subject: Re: [nod-ai/SHARK-Turbine] torch.aten.reflection_pad1d (Issue #293) Caution: This message originated from an External Source. Use proper caution when opening attachments, clicking links, or responding. @kumardeepakamd<https://github.com/kumardeepakamd> / @frederik-h<https://github.com/frederik-h> can you coordinate? @kumardeepakamd<https://github.com/kumardeepakamd> I have created a draft PR for the 2d operator: llvm/torch-mlir#2693<llvm/torch-mlir#2693>. I think the 1d operator is just a special case, i.e. it should probably be converted to the 2d operator, but it will still require separate testing etc. — Reply to this email directly, view it on GitHub<#293 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/A5OMX3ZTRLLIZOYHUSPQBVTYKWMLVAVCNFSM6AAAAABA6WCOUSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQNRXG43TCMBRHA>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[frederik-h]

@kumardeepakamd Is it ok to close this issue since llvm/torch-mlir#2706 has been merged? Or is there anything else left to do?