search

intel / torch-xpu-ops

Apache License 2.0
25 stars 18 forks source link

[E2E][Triton] TIMM model pnasnet5large accuracy regression. #408

Open fengyuan14 opened 3 months ago

fengyuan14 commented 3 months ago

🐛 Describe the bug

Skip the model in pre-ci. Please retrieve the case, if triton gets fixing. https://github.com/intel/intel-xpu-backend-for-triton/issues/1353

Versions

Latest main.

etaf commented 3 months ago

This issue have been fixed in latest triton main branch. we'll update triton to resolve this issue.

chuanqi129 commented 2 months ago

@mengfei25 please help to double check whether this issue got fixed

etaf commented 2 months ago

Please keep this open issue, we've not updated triton.