BradPepersAMD

Results 24 comments of BradPepersAMD

Implement SigmoidFocalLoss operation

MIOpen is moving to the new monorepo setup and all older unmerged PR's are being closed. Please re-open this as part of the new repo if these changes are still...

Implement MultilabelSoftMarginLoss for small C

MIOpen is moving to the new monorepo setup and all older unmerged PR's are being closed. Please re-open this as part of the new repo if these changes are still...

Implement Cumulative reduction (max, min, sum, prod) forward with small last dim

MIOpen is moving to the new monorepo setup and all older unmerged PR's are being closed. Please re-open this as part of the new repo if these changes are still...

[gTest] Add more test coverage for Batchnorm Backward

MIOpen is moving to the new monorepo setup and all older unmerged PR's are being closed. Please re-open this as part of the new repo if these changes are still...

MIOpen error tolerance

Did this pass CI and are we good to merge this now?

Forbid inefficient TensorDescriptor initialization

MIOpen is moving to the new monorepo setup and all older unmerged PR's are being closed. Please re-open this as part of the new repo if these changes are still...

[MHA] Implement ExecutionPlan execute function

MIOpen is moving to the monorepo and old PR's need to be closed for this to happen. If needed, this PR can be re-opened in the new repo.

[Conv] add ck bfp16 for fwd and bwd

What about WRW? Is it already enabled for BF16?

[CONV] fix naive conv kernel for large tensors

This still needs to be fixed but is now stale so closing.

rjs/rk gtc

MIOpen is moving to the new monorepo setup and all older unmerged PR's are being closed. Please re-open this as part of the new repo if these changes are still...