Ted Themistokleous

Hi @snnn, the MIGraphX side has completed our review process for the changeset for this and I just pushed up a change the other day that shouldn't break functionality between...

Force pushed after a rebase off your main branch. Looks like I had a conflict with another upstream change that broke CI. Was able to reproduce and fix locally. Not...

> So `SoftmaxCrossEntropyLoss` is a onnx function which means it can be implemented using a composite of other onnx operators(like `DynamicQuantizeLinear`). So we shouldn't need to add a new operator...

> > > Furthermore `NegativeLogLikelihoodLoss` is also an ONNX function as well, so we dont need to write a custom operator for this as well, although I couldn't find a...

Rebasing this off master to restart work on this tomorrow

Need to fix parser tests but verify tests all pass and have been validated using the gold data. Gold data check leveraged Onnxruntime and numpy to ensure our answers are...

> Is there any verify test which compares ref value with the equivalent gpu based output? Thanks. Not needed here as this is an onnx function which is made up...

> > > Is there any verify test which compares ref value with the equivalent gpu based output? Thanks. > > > > > > Not needed here as this...

Got all the end to end stuff working, seeing some odd behavior with the dnn1 based tests for double/float. Looks like their version of log doesn't support double which causes...

Use the following target models for testing: resnet50 Bert distilgpt2 Babystep the process for this and see what we need/can leverage from existing backends/Execution provider - [x] Get triton installed...