James Newling

Maybe extending the logic in `AtenWhereSelfOp::fold` ?

This doesn't look correct to me, I think we changed to import torchscript recently https://github.com/llvm/torch-mlir/pull/2842

Thanks @VimalWill @saienduri can you please for advise?

``` :929:10: error: failed to legalize operation 'torch.aten.view' that was explicitly marked illegal %7 = torch.aten.view %5, %6 : !torch.vtensor, !torch.list -> !torch.vtensor ``` should this be handled by torch.aten.view...

To reproduce this error ```python import torch from torch_mlir import torchscript class OnesTrilViewModule(torch.nn.Module): def __init__(self): super().__init__() def forward(self, x): key_size = x.shape[0] return torch.tril(torch.ones((key_size, key_size), dtype=torch.bool)).view( 1, 1, key_size, key_size...

I'm imagining a utility function which attempts to get a symbolic shape for any value. ```FailureOr getSymbolicShape(Value v);``` The implementation recursively walks through producers until it has something to return,...

I would like to load onnx models, and run them (on CPU). Mostly for testing: I would like to compare the outputs obtained with my (C++) implementations. Not urgent.

Apologies for the slow reply. You can build poprithms with the cmake flag`POPRITHMS_BUILD_MOCKS` set to `OFF` as described at https://github.com/graphcore/poprithms

Hi Jack, If you want to make MIOpenGEMM run fast for a particular Geometry (say M=10, N=5, K=4096, tA=tB=0), take a look at: `examples/find.cpp`. If you want to actually see...

These one of kernel performance tests are tricky to get right in my experience. One thing to note is that the first call to gemm with a Geometry (M=10, N=4096,K=25088,TransA=false,...