Skip to content
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.

Sign up for GitHub

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

Jump to bottom

TIMM unsupported ops #36

Open
5 of 9 tasks
shivadbhavsar opened this issue Sep 14, 2023 · 5 comments · Fixed by #46
Open
5 of 9 tasks

TIMM unsupported ops #36

shivadbhavsar opened this issue Sep 14, 2023 · 5 comments · Fixed by #46
Assignees
@TedThemistokleous
Labels
TorchMIGraphX For internal tracking

Comments

@shivadbhavsar
Copy link
Collaborator

shivadbhavsar commented Sep 14, 2023 •
edited by TedThemistokleous
Loading

  • aten.stack.default (crossvit_9_240)
  • aten.elu.default (dpn107)
  • aten.unfold.default (eca_halonext26ts)
  • aten.hardswish.default (fbnetv3_b, lcnet_050, levit_128, mobilenetv3_large_100)
  • aten.hardtanh.default (mobilenetv2_100, rexnet_100)
  • aten.roll.default (swin_base_patch4_window7_224)
  • aten.im2col.default (tnt_s_patch16_224, volo_d1_224)
  • aten.max.dim (volo_d1_224)
  • aten.col2im.default (volo_d1_224)
@shivadbhavsar shivadbhavsar added the TorchMIGraphX For internal tracking label Sep 14, 2023
@TedThemistokleous TedThemistokleous linked a pull request Oct 25, 2023 that will close this issue
Update aten operators #46
Merged
@TedThemistokleous
Copy link
Contributor

hardtan and hardswish completed and in PR

@TedThemistokleous
Copy link
Contributor

Some issues with max due to return args. Moving this to other sprint. Will tackle stack operator next.

Leaving estimate the same for now

@TedThemistokleous
Copy link
Contributor

TedThemistokleous commented Nov 10, 2023 •
edited
Loading

Stack complete. working on the other ops, meeting with Shiv this afternoon to sort out issue we're seeing with stack.

Adding min since its the similar operator but same structure/test procedure as max and leverages MIGraphX's reduce_op. This completes coverage for the reduce_op set (max/min/sum/mean) once completed.

@TedThemistokleous
Copy link
Contributor

Pushed up fix for max/min. Added argmin to support min as well. Similar to argmax

@TedThemistokleous
Copy link
Contributor

Reopening for the last 4 ops that are needed

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@TedThemistokleous TedThemistokleous

Labels
TorchMIGraphX For internal tracking
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

Update aten operators ROCm/torch_migraphx
2 participants
@shivadbhavsar @TedThemistokleous